## Of jackal and hide and Old Sinitic reconstructions

[The first page of this post is a guest contribution by Chris Button.]

I've been thinking a little about the word represented by chái 豺* which I would normally reconstruct as *dzrəɣ (Zhengzhang *zrɯ) ignoring any type a/b distinctions. However, it occurred to me that a reconstruction of *dzrəl (for which Zhengzhang would presumably have *zrɯl) would give the same Middle Chinese reflex (I'm not citing Baxter/Sagart since they don't support lateral codas presumably for reasons of symmetry). I'm not sure if outside of its phonetic speller cái 才 there is any reason to go with -ɣ rather than -l in coda position for 豺. However, if we go with a lateral coda as *dzrəl, it looks suspiciously similar to Old Iranian šagāl from Sanskrit śṛgāla (perhaps even more so if we fricativize the Old Iranian /g/ to /ɣ/ intervocalically as in modern Persian).

[*VHM:  This is always a challenging word for translators.  "jackal" and "dhole" are two possibilities.]

What do Language Log readers think of Chris's postulations?

A brief note on the etymology of English "jackal" from the American Heritage Dictionary, 5th ed.:

[Turkish çakal (influenced in English by Jack, man, fellow), from Persian šaghāl, from Middle Indic sigāl, from Sanskrit śṛgālaḥ, of unknown origin.]

Let me [VHM] begin my part of this joint post with a brief explanation of how it happened.  To put it simply, last night as I was walking home, I was trying to think of a title for Chris's post, and the first thing that popped into my mind (don't ask me why, that's just the way minds work), was "Jackal and hide".  As I continued walking, I started to wonder about Sino-IE possibilities for "hide", since Sino-IE interactions are the theme of Chris's note on "jackal".  By the time I reached the door of my house five minutes later, I had sketched out the rudiments of what follows.

The Sinitic word for "skin; hide; pelt; fell" [N.B.!] is pí 皮, Middle Sinitic (MS) /bˠiᴇ/ (Zhengzhang) and Old Sinitic (OS) /*bral/ (Zhengzhang).

I remember, more than three decades ago, reading through the whole of Carl Darling Buck's magisterial A Dictionary of Selected Synonyms in the Principal Indo-European Languages (Chicago:  University of Chicago Press, 1949) searching for groups of cognates of distinctive cultural terms that were similar in Old Sinitic and Indo-European languages.  Many of the groups I identified at that time have continued to stick in my mind all this while, and they have proven useful in the series of posts about Sino-IE correspondences I've written on Language Log for the last couple of years and elsewhere for a much longer period of time.

One such group that stayed strongly in my mind for all these years was Buck, 4.12 (p. 200b), namely Latin pellis, with cognates in Italian, French, Romanian, Spanish, Gothic, Old Norse, Old English, Middle English, New English, Dutch, Old High German, Middle High German, New High German, Greek, and Lithuanian.  There are no secure cognates of the *pel ("skin") root in Tocharian.

Thus the Latin word for "hide; skin" is pellis, and there are cognates in many European languages, including even in English ("fell" [N.B.!] and "pelt").  For updating and ease of transfer, here is the relevant section from the American Heritage Dictionary of Indo-European roots:

## pel-3

Skin, hide.

1. Suffixed form *pel-no-. fell3 from Old English fell, skin, hide, from Germanic *felnam.
2. film from Old English filmen, membrane, from Germanic suffixed form *fel-man-ja-.
3. Suffixed form *pel-ni-. pelisse, pellicle, pelt1, peltry, pillion; pellagra, surplice from Latin pellis, skin.
4. erysipelas from Greek -pelas, skin.
5. Suffixed form *pel-to-. peltate from Greek peltē, a shield (made of hide).

[Pokorny 3b. pel- 803.]

The last item rings resonant bells, does it not?  See "Of armaments and Old Sinitic reconstructions, part 6" (12/23/17), where we studied words for that type of shield in considerable depth and breadth and found persuasive evidence for correlations between Graeco-Latinate and Sinitic terminology.  So what we said there about the particular type of shield called a peltē in Greek and fá 瞂 (MS bjwot) in Sinitic mutually reinforces what we have discovered here about parallels between Graeco-Latin, Germanic, and Lithuanian cognates for "skin; hide; pelt; fell" and Sinitic pí 皮, Middle Sinitic (MS) /bˠiᴇ/ (Zhengzhang) and Old Sinitic (OS) /*bral/ (Zhengzhang).

The following, concluding section is relevant both for Chris's part of this post and mine.

Skeptics will naturally want to question, even if there are strong correlations between particular cognate sets of terms in IE and Sinitic, how could they possibly come in contact in premodern times?  The answer is very easy:  based on archeological, anthropological, historical, art historical, technological, cultural, genetic, and linguistic evidence, there have been robust contact and exchange between the eastern and western portions of Eurasia (bronze, wheat, horses, chariots, weapons, textiles, magi, and so forth) for at least the last four millennia and more.

What is especially noteworthy is that there is clear evidence for the involvement of numerous IE groups in the eastward transmission of cultural attributes, including language:  Greek, Latin, Tocharian, Indic, Iranian, Germanic, etc.  See, among other sources, the following works:

Victor H. Mair, ed., The Bronze Age and Early Iron Age Peoples of Eastern Central Asia (Washington, D.C.: Institute for the Study of Man Inc. in collaboration with the University of Pennsylvania Museum Publications, 1998).  2 vols.

P. Mallory and Victor H.Mair,The Tarim Mummies: Ancient China and the Mystery of the Earliest Peoples from the West (London:  Thames & Hudson, 2000).

"Early Indo-Europeans in Xinjiang" (11/19/08).

Victor H. Mair, ed., Contact and Exchange in the Ancient World (Honolulu:  University of Hawai'i Press, 2006).

Victor H. Mair, "Language and Script: Biology, Archaeology, and (Pre)History," International Review of Chinese Linguistics, 1.1 (1996), 31a-41b.

Mair, Victor H., "Old Sinitic *Myag, Old Persian Maguš, and English Magician", Early China, 15 (1990): 27–47.  Available on JSTOR here.

Plus all of the previous posts in this long-running Language Log series on IE and Old Sinitic reconstructions

A final word:  with their horses, carts, wagons, and chariots, IE peoples were highly mobile from the fourth millennium BP on.  Just taking the Scythians as an example, already from about the 8th century BC, they were ranging from western Eurasia through Central Asia to East Asia and even down into Southeast Asia, taking with them their technology and art.

As for the proto-Germanic peoples, they may have been active as a linguistic entity by around 500 BC, and the Greeks were ensconced in Central Asia by at least the latter part of the 4th century BC, having a strong impact on the local cultures and beyond, all the way to the East Asian Heartland (EAH).  Alexander the Great's generals founded walled cities as far east as Alexandria Eschate ( Ἀλεξάνδρεια Ἐσχάτη ["Alexandria the Farthest"]) at the southwestern end of the Ferghana Valley in modern Tajikistan.  See "The Greco-Chinese War Over the Heavenly Horses".  The research of Richard Barnhart (Yale) and Lukas Nickel (Vienna) has revealed evidence of Greek influence on the terracotta army of the First Emperor of the Qin (259-210 BC).  Other scholars have demonstrated traces from the west through the investigation of beads, glass, coins, and so on already during BC times.  Linguistically, as we have shown before on Language Log and elsewhere, the words for "coral", "honey", and a famous legendary sword come from Iranian (Khotanese Saka), Tocharian (a well-known borrowing), and Germanic (see the very first post in this series [3/8/16]) — all from Han Dynasty times (202 BC-220 AD) or earlier.

[Thanks to Hannes Fellner, Douglas Adams, and Lucas Christopoulos]

1. ### Victor Mair said,

December 16, 2018 @ 9:59 am

I forgot to mention Tsung-tung Chang's well-known "Indo-European Vocabulary in Old Chinese" which was published as Sino-Platonic Papers, 7 (January, 1988), i + 56 pp. (free pdf)

On p. 21, Chang notes the Graeco-Latin group of cognate words meaning "fell; skin" and compares it to Old Sinitic beia 皮.

2. ### Victor Mair said,

December 16, 2018 @ 11:51 am

From Don Ringe:

No obvious Tocharian cognates for pellis, etc. TB pilta 'leaf' is conceivable, but the semantics are unimpressive; more likely it's not connected. TB pile, TA päl 'wound' is certainly not connected; they go with Gk. ápelos 'wound', PIE *h2pélos, with a laryngeal-initial root.

It is difficult to point to a time when PGmc might have been spoken. PGmc. seems to be the Jastorf culture in southern Denmark and adjacent areas to the south–or rather part of it; early Jastorf (ca. 750 BCE) is too early, and late Jastorf and successor cultures (ca. 250) covers too broad an area for a single language. Corollary: the tribes identified as Germans by the Romans probably did *not* speak our reconstructed PGmc. (mostly), but closely related sister dialects/languages. But between early Jastorf, which is clearly pre-PGmc., and the last common ancestor with any other surviving group (probably pre-Proto-Balto-Slavic) there is a wide gulf of time, possibly a couple thousand years. Strictly speaking, any generation in there is a generation when pre-PGmc. was spoken, but the further back you go the less like our reconstructed PGmc. it would be. I guess I'd have to say, it depends on what you want to try to do with it.

3. ### David Marjanović said,

December 16, 2018 @ 2:50 pm

OK, let's accept */bral/ as the Old Sinitic form of 皮 (though I'd love if anyone could find the Baxter&Sagart version of this). Let's further blame the */e/-*/a/ mismatch on Indo-Iranian or something. That still leaves us with two questions: where does the */r/ come from, and where does the voice of the */b/ come from?

Corollary: the tribes identified as Germans by the Romans probably did *not* speak our reconstructed PGmc. (mostly), but closely related sister dialects/languages.

In any case, some of the last sound shifts that must have happened before the Proto-Germanic stage were still absent in forms recorded by Caesar and even Tacitus. Off the top of my head there are the Fenni, lacking the */enC/ > */inC/ shift (C = any consonant, including "another" */n/) that is present in the Φιννοί of Ptolemaeus and all later sources up to the present day, and Segimer, the father of Arminius, who lacks the */e…i/ > */i…i/ umlaut phenomenon that has given us the modern name Si(e)gmar. Either the Proto-Germanic stage was spoken between Tacitus in the 1st and Ptolemaeus in the 2nd century, or Tacitus only had access to dialects that were very closely related to but not descended from Proto-Germanic.

4. ### David Marjanović said,

December 16, 2018 @ 3:38 pm

Linguistically, as we have shown before on Language Log and elsewhere, the words for "coral", "honey", and a famous legendary sword come from Iranian (Khotanese Saka), Tocharian (a well-known borrowing), and Germanic (see the very first post in this series [3/8/16]) — all from Han Dynasty times (202 BC-220 AD) or earlier.

I have just reread the sword posts and their discussion threads. The Germanic and Abkhaz-Abaza words are pretty similar (though the latter is more similar to the Northwest Germanic form than the expected East Germanic one, while the Finnic loan must have been from East Germanic and the Slavic one at least cannot have been from Northwest Germanic). But between the northwestern tip of the Caucasus and China, we found nothing…?

5. ### Victor Mair said,

December 16, 2018 @ 4:39 pm

Well, how did Tocharian, the second oldest IE language, and centum to boot, get where it was on the back door of Sinitic?

6. ### DMT said,

December 16, 2018 @ 9:30 pm

There is a pdf of Baxter-Sagart reconstructions available online; it gives *m-[p](r)aj for 皮.

7. ### Chris Button said,

December 16, 2018 @ 11:06 pm

A medial /r/ in 皮 isn't required which is why Baxter & Sagart place it in parentheses as an optional component. Following Pulleyblank's system of OC rhymes, I would go with *bàl (the grave accent indicating a Type-B syllable). The difficulty with Baxter & Sagart's -j coda here is that it requires a rather arbitrary (at least in my opinion) rule, described by Baxter (1992) as "-aj monophthongization", to account for the coda's demise by the time of Middle Chinese. Conversely, a separate (equally arbitrary in my opinion) rule of "j-insertion" is postulated by Baxter to account for his reconstruction of cases like 豺 with an open rhyme as *-ə where Pulleyblank instead has *-əɣ (which he later treats as *-əɰ to specify the coda as an approximant).

8. ### Chris Button said,

December 17, 2018 @ 12:18 am

The difficulty with Baxter & Sagart's -j coda here is that it requires a rather arbitrary (at least in my opinion) rule, described by Baxter (1992) as "-aj monophthongization", to account for the coda's demise by the time of Middle Chinese.

Before I'm accused of misrepresenting anything, I should probably add that Pulleyblank was not actually against such a "rule" since his *-al essentially represents a prior stage to Baxter's *-aj on the basis of evidence for a *-j coda in Min dialects and Vietnamese and Korean loans. However, my hunch (quite possibly incorrect) is that this was rather due to the affinity of -l and -j, as clearly attested by the well-known confusion of the *-əl and *-əj rhymes in Pulleyblank's reconstruction (Baxter's *-əj and *-i(j) in which the latter is highly problematic once one steps out of the realms of abstract phonology), than anything else (i.e. *-l could be represented by -j in such cases without necessarily having shifted in such a way during its evolution into Middle Chinese). With that being said, Pulleyblank's *-l was rather designed to account for contacts with *-n where Baxter & Sagart now reconstruct **-r to give *-n or *-j dialectally. Personally, as much as I respect Baxter & Sagart's work (in spite of often disagreeing with many of the conclusions it reaches), I'm not currently convinced by most of what they attribute to different "dialects".

9. ### martin schwartz said,

December 17, 2018 @ 2:12 am

@chris button:
$=sh, G = gamma, th = theta, ' = 'ay(i)n, tx = tsh: No such word for 'jackal' in Old Iranian (Avestan and Old Persian). New Persian has$aGâl 'jackal',

which somehow goes with the Skt. (itself obscure etymologically), maybe via Middle Indic. Man. Sogd has $Vkârê 'jackal'; OIr *g would give Sogd. 'G, so prob Sogd. < Indic. Possible roaming wild dog words or coincidents: Semitic root THETA-'A(Y)IN-L in Heb.$u'âl 'fox' and Arabic and Aram. cognates.

Georgian dzaGli and Basque txakur 'dog'. Hey, are we back to dogs again?
I'll pass over the reconstruction of 'lion' in Chinese, Sogd. $argu-,Khwar. sargu-, etc., and Skt. /sinha-/, which I remember from Boodberg and Henning. 10. ### Martin Schwartz said, December 17, 2018 @ 2:22 am for the computer-garbled line: Man. Sogd has$Vkarie jackal' … < Indic.

Roaming or coincidentally wild dogwords:
Heb. $u'âl 'fox', etc. 12. ### martin schwartz said, December 17, 2018 @ 2:30 am for some reason, my dollar sign = sh gets garbled. The Heb. sh reflects proto-sem. theta. 13. ### Theophylact said, December 17, 2018 @ 12:00 pm I've read this far and still haven't come up with an excuse for a "Jekyll and Hyde" pun. I'm disappointed with myself. 14. ### Aidan said, December 17, 2018 @ 1:38 pm That reminds me of Tibetan spyang, as in spyang-ki, "wolf, jackal" and 'phar-spyang: "dhole"; but I'm afraid I don't know Sino-Tibetan sound correspondences well enough to evaluate if spyang and *dzrəɣ(l) look related or not. 15. ### Chris Button said, December 17, 2018 @ 10:15 pm @ Martin Schwartz No such word for 'jackal' in Old Iranian (Avestan and Old Persian). New Persian has$aGâl 'jackal', which somehow goes with the Skt.
(itself obscure etymologically), maybe via Middle Indic. Man. Sogd has $Vkârê 'jackal'; OIr *g would give Sogd. 'G, so prob Sogd. < Indic. Thanks for the clarification. It's good to see something in Sogdian. By the way, Jim Matisoff has an article "Toward a Eurasian Bestiary" (2009) where he talks about "Jackal" and adds the following note: "Medieval and modern Indo-Aryan forms include Prakrit si(g)āla-, Kashmiri śāl, Nepali siyāl, Bengali siyāl, Assamese xiyāl. In recent times several TB languages of India and the Himalayas have reborrowed this word from Indo-Aryan. These back-loans include Limbu syaʔl (< Nepali), Kanauri shyáles ~ shyálí (prob. < Kashmir), Mizo shihal (< Assamese), and Mikir hijai (prob. also < Assamese)." Georgian dzaGli and Basque txakur 'dog'. Hey, are we back to dogs again? The Georgian form apparently goes back to Proto Kartvelian *ʒ₁aɣl- which is also suspiciously close to 豺 *dzrə́l https://archive.org/stream/KlimovEtymological/Klimov#page/n299/mode/2up In his Handbook of PTB, Matisoff tentatively suggests an association of the "jackal" forms with 犬 *kʰʷə́ɲʔ and its PIE association that we discussed in "Of Dogs and Old Sinitic" linked above, but I'm not convinced by the phonology. Possible roaming wild dog words or coincidents: Semitic root THETA-'A(Y)IN-L in Heb.$u'âl 'fox' and Arabic and Aram. cognates.

Not sure about that one. However, speaking of Hebrew, the notion that the phonetic 才 in 豺 may reflect later phonological shifts is paralleled by 桂 "cinnamon, cassia" which I would normally reconstruct as *qájs (its phonetic is 圭 *qáj) but whose evolution seems more likely to be from earlier *qjáts via an association with Hebrew qetsia "cassia".

@ Aidan

That reminds me of Tibetan spyang, as in spyang-ki

Not sure about that one either, although Matisoff (2009) suggests the "ki" component to be from the same "dog" root with which we can ultimately connect 犬 *kʰʷə́ɲʔ. I actually contributed a brief discussion of the denasalization of the OC coda in PTB specifically in terms of Burmese to the body of Prof. Mairs earlier "Of dogs…" post.

16. ### Victor Mair said,

December 18, 2018 @ 8:24 am

From Gerd Carling:

The PIE root *pel- is a very complex and rich tree with various forms, found in many IE languages, but not in Tocharian. The Toch B word for leather is ewe 'inner skin, hide, leather' (likely from IE *h1euH- 'put on clothes') or yetse 'skin (person's)' (Toch A yats) (these words have no reliable etymology); AB have no words for 'fur' or 'leather' (first meaning) as I am aware of, which is mysterious.

17. ### Victor Mair said,

December 18, 2018 @ 8:40 am

From Melanie Malzahn:

The underlying root is only attested in n-stem nominal forms meaning 'skin' (i.e. not independently) and precisely in the languages you mentioned. No cognates are (so far) attested in Tocharian.

I'd say the Germanic forms are inherited from PIE and not proto-Germanic neoformations.

18. ### Chris Button said,

December 19, 2018 @ 6:21 am

Conversely, a separate (equally arbitrary in my opinion) rule of "j-insertion" is postulated by Baxter to account for his reconstruction of cases like 豺 with an open rhyme as *-ə where Pulleyblank instead has *-əɣ (which he later treats as *-əɰ to specify the coda as an approximant).

As a random aside, outside of the extensive internal support in Old Chinese, Pulleyblank's velar coda *-ɣ/ɰ also offers an interesting solution to the long-standing question about whether 絲 (糸) *sə̀ɣ "silk" is related to Greek sḗr (from whence Latin Sēres "Chinese/China").

Without the -ɣ coda in OC, it is difficult to account for the Greek trilled -r coda, but an association of rhotics with back articulations is well-known (e.g. the shift in French of trilled /r/ to uvular /ʁ/). The only issue is that the direction of change would normally be expected to be from /r/ to /ɣ/ as if the word entered Chinese from Greek. However, the fact that Pulleyblank treats /ɣ/ as an approximant /ɰ/ suggests that the OC coda was simply associated with the most appropriate sound in the Greek phonological inventory.

19. ### David Marjanović said,

December 23, 2018 @ 5:26 pm

Well, how did Tocharian, the second oldest IE language, and centum to boot, get where it was on the back door of Sinitic?

For this case we have at least an archeological scenario, the Afanasievo culture and its successors. For Germanic we have nothing that would indicate contacts or movements (through Balto-Slavic and Iranian) that far east.

BTW, if (as I agree) Tocharian is "the second oldest IE language" ( = the sister-group to all of IE except Anatolian), then the fact that it's centum ( = has merged the palatalized velars into the plain ones) must be a coincidence, because Proto-Rest-IE obviously lacked both the kentum and the satəm merger.

20. ### David Marjanović said,

December 23, 2018 @ 5:30 pm

Oh, I forgot:

However, the fact that Pulleyblank treats /ɣ/ as an approximant /ɰ/ suggests that the OC coda was simply associated with the most appropriate sound in the Greek phonological inventory.

That would work with an English-style [ɹ], but not for a trill. I'm sure any Ancient Greek exposed to [əɰ] would either not hear the [ɰ] at all, or would hear the sequence as a diphthong that would end up equated with… pretty much any of the existing Greek diphthongs, I guess.

Aren't there transcriptions where characters with a Baxter/Sagart *-r are used to represent foreign [r]?

21. ### Chris Button said,

December 23, 2018 @ 11:28 pm

Well don't forget the uvular trill /ʀ/ between the alveolar trilled /r/ and uvular fricative /ʁ/ stage in French. Even if we go with Pulleyblank's original /ɣ/ rather than /ɰ/ (as I actually tend toward in any case), an association of the vibration of /r/ with the frication of /ɣ/ is surely not entirely untoward?

22. ### David Marjanović said,

December 25, 2018 @ 4:55 pm

Sure, I use [ʀ] natively and have misheard foreign uvular approximants as that. But someone who isn't familiar with it will not hear [ɰ] or [ɣ] as their native [r].

Naturally, given the great distance, the word could have passed through any number of languages between Greece and China. But which ones have a [ʀ], or have a /ʁ/ but no /r/?