Language Log

« previous post | next post »

[This is a guest post by Rhona Fenwick]

Though best-known for its titanic consonantal inventory, Ubykh also has an etymologically fascinating vocabulary, heavy with loans from a diverse array of sources. Many of these are drawn from the indigenous lexicons of its Circassian and Abkhaz sisters, but Circassian and Abkhaz both also acted as proxies by which Ubykh became a linguistic placer deposit of sorts, receiving substantial loan strata from millennia of the ebb and flow of Kartvelian, Turkic, Mongolic, Semitic, and Indo-European cultural tides. More recently the Ubykh nation’s exodus from their homeland and subsequent exile in Anatolia, following extensive genocides at the end of the Great Caucasian War (Ubykh: Adəɣaʁʷərda ‘the Rape of Circassia’), added yet another layer of complexity and invested the language with loans from whole new branches of Indo-European, Turkic, and Semitic. This makes compiling an Ubykh etymological dictionary a complex and challenging project, and while engaged in it I’ve often found myself having to track etymologies along paths that lead deep into other language stocks entirely. This post began as a question to Victor Mair while I was playing bloodhound along one such trail, and it was on his suggestion that I reworked it into a post for LL. Thank you for having me!

The trail I’ve been following is in the etymology of the Ubykh word zənʤə́r ‘chain’, a quite recent loan alongside the indigenous, more usual term kʷələ́ (whose etymology, incidentally, remains unclear). It’s not certain whether Ubykh zənʤə́r is a direct borrowing from Ottoman Turkish zincir, or was introduced from Ottoman via the intermediary of Abkhaz азынџы́р (a-)zənʤə́r, but in any case the Ottoman term itself has been in use since the Old Oghuz / Old Anatolian Turkish period (Old Oghuz zencīr), and thence arises straightforwardly from Classical Persian zanǰir ~ zenǰir ‘chain’. This in turn comes from Middle Persian zanǰīr (Pahlavi spelling zncyl), with clear cognates also in Parthian zenčīr (Parthian spelling zyncyhr, where the great Iranist Walter Henning believed –h– is probably (p.240) non-etymological) as well as Buddhist Sogdian zyncry’kh and Manichaean Sogdian jyncry’.

A priori, it’s tempting to view these as compounds from a Middle Iranian root *zen– or *zin– ‘tie, bond, fetter’, perhaps supported by further Sogdian forms such as zynγwd ‘harness’, zyncyk ‘harnessed’, and a series of possible figurative derivatives referring to more abstract forms of bondage (if one can adduce zyn’y ‘deposit, trust, care’, zynβrt ‘foster-parents’, zynβr’nch ‘trusteeship’). But *zen– ~ *zin– doesn’t appear to be reconstructible outside Sogdian with this sense. Some have tried to connect this form with the Middle Iranian *zēn– seen also Middle Persian zēndān, Parthian zēnδān ‘prison’, but this idea has been mostly abandoned: more likely in these terms is a semantic shift from an earlier *‘arsenal, armoury’, with the Middle Iranian derivational suffix –dān ‘holder, container’ applied to zēn ‘weapon, armour’, descended from Proto-Iranian *zaina– ‘weapon’. (The stem is notably reflected in Avestan zaēna– ‘weapon’, but reflexes have also furnished widespread loans, including Old Armenian zēn ‘weapon’, Syriac, Hebrew zaynā’ ‘id.’, and Tocharian B tsain ‘arrow’.) The etymological path for ‘chain’ has usually ended here in Middle Iranian for this reason.

While looking for Indic parallels in Monier-Williams’s vast 1899 dictionary, though, I came across a striking comparandum in Sanskrit hiñjīra– (m.) ‘rope or chain for fastening an elephant’s foot’. This term, though uncommon, is an essentially perfect cognate for the Middle Iranian forms, right down to the regular inherited Sanskrit / Iranian h-/z– correspondence (from Proto-Indo-Iranian *j́ʰ-, usually from PIE *ǵʰ-). From the Sanskrit and Middle Iranian terms, then, one can reconstruct a robust Proto-Indo-Iranian *j́ʰinǰīra– ‘chain, fetter’. This veers sharply away from Indo-European for two reasons:

1. If the Middle Iranian ‘chain’ words were in fact from Proto-Iranian *zaina– (or even a zero-grade *zina-), they should have corresponded to Proto-Indo-Iranian **j́ʰainaǰīra– ~ **j́ʰinaǰīra– and thence Sanskrit *henajīra– ~ *hinajīra-. Thematic vowels formed an integral part of compounding stems until well after the breakup of Indo-Iranian, as seen in the oldest stages of both branches (compare, for example, Sanskrit úṣṭṛa-sthāna– = Avestan uštro-stāna– ‘camel stall’, Old Persian uša-bāri– ‘camel-borne’).

2. More problematically, *j́ʰinǰīra– really can’t reflect any plausible PIE nominal form. It’s clearly not a root noun, for several reasons (two syllabic nuclei, no legal position for an ablauting vowel, apparent primary long *-ī-…). Even if one treats the form as derived, taking the long *-ī– rather as *-ih_X– and segmenting out suffixes to isolate a plausible root—PIE **ǵʰing-ih₂-ro– would in principle be most likely—the necessary PIE suffix combination is morphologically contradictory: *-ih₂ is a feminine derivational suffix of nouns (and, moreover, is ablauting: its full grade is *-yeh₂, which one would expect to show up with a zero-grade root), but the two common suffixes of the form *-ro– are the Caland stem-formant –ró-, which attaches rather to adjectival roots, and the nominaliser –ro-, which attaches to verb roots. And in any case, the root left behind (**ǵʰyeng-?) has no known comparandum either in Indo-Iranian, or anywhere else in PIE.

In view of this, the Proto-Indo-Iranian form itself is more likely a loan, which sent me off looking elsewhere. And it turns out that a similar form does exist elsewhere, in Written Mongol ginǰi ‘chains, fetters, shackles’. This is likely not a mere Iranian borrowing, though: Mongol z– or ǰ– would have been expected rather than g– in a Persian, Parthian, or Sogdian loan, and Written Mongol also quite happily tolerates final –r, preserving it regularly in other Persian loans. (Compare Written Mongol balur ‘crystal’, siqir ‘sugar’, debter ‘booklet, register’ alongside their Middle Persian equivalents bēlūr, šakar, daftar; though this last seems to have entered Mongol through Tibetan deb ther, still the final –r is preserved. In fact, I haven’t yet found a loan from any language where Written Mongol has deleted final –r, though I’m still looking.)

Then, having read Victor’s recent post pondering a possible link between Proto-Indo-Iranian *úštras "camel" and Sinitic 駱駝, a new angle of enquiry occurred to me: given that Proto-Indo-Iranian *j́ʰinǰīra– appears to be non-IE in form, and on the assumption that Written Mongol ginǰi isn’t a Middle Iranian loan, I wondered if these forms—each one treatable as two CVx syllables with obstruent onsets and closed by sonorants or zero—might reflect an old disyllabic Sinitic compound, borrowed separately into Mongolic and Indo-Iranian. While I’ve been able to wrap my head around most other historical linguistic literatures I’ve needed so far, Sinitic is a uniquely complex field and one that resists simple skimming by a Sinitically-untrained linguist like myself; all I was able to come up with (after encountering modern Mandarin liànzi ‘chain’) was to wonder whether the second syllable of the Mongol and Indo-Iranian terms might arise from the Sinitic noun-formant 子. It was at this time that I reached out to Victor, who very kindly supplied some relevant information on that point:

Jerry Norman (Chinese, p. 114) believed that 子 did not become fully grammaticalized as a noun formative for all types of nouns until Tang times (618-907). Before that, it served as a diminutive suffix, sometimes with pejorative implications, for a rather long period. However, it can be found as a plain noun suffix on the names of animals and tools in texts of the Nanbeichao (Northern and Southern dynasties) period (420-589). Norman's primary sources for information of this type were Ōta Tatsuo / Ohta Tatsuo’s works on historical lexicon and grammar, with some input from Wang Li’s lexicon volume from his history of Chinese.

Since Proto-Indo-Iranian (whose breakup is conventionally dated to the first half of the second millennium BC) is chronologically well out of reach of the Nanbeichao era, though, unfortunately 子 becomes unlikely as a source for the second element of *j́ʰinǰīra-, unless the Indo-Iranian and Mongol terms arise from a much earlier Sinitic diminutive of some sort (which remains at least plausible, though would depend on the nature of the source term). So I remain for the moment with Proto-Indo-Iranian *j́ʰinǰīra– in one hand, Written Mongol ginǰi in the other, and a gut feeling that some missing link connects the two. Perhaps one of my good colleagues here at Language Log might know what it is.

January 27, 2021 @ 10:08 am · Filed by Victor Mair under Borrowing, Etymology, Historical linguistics, Reconstructions

Permalink