Inexplicably unproposed Uralic etymologies

There are resemblances between some Mari words and items in other Uralic languages that are extremely blatant and yet, to my knowledge, have gone uncommented. That’s not to say that an etymological link is tenable, but one would expect the UEW or other general references to at least note and shoot down some prior attempt to relate the given words. Why have the following not been compared before, even in the heady early 20th century when standards were relatively lax?

  • Finnish salama ‘lightning’ ~ MariE šolem ‘hail’. Yes, there’s a difference in meaning, but it’s not unusual for words denoting weather conditions to shift semantically, e.g. MariE jür ‘rain’ < Cv. yur ‘snow’. I suppose the difficulty here is that MariW šolem doesn’t show /a/ as some might have earlier expected in a word from *salama. However, it does agree with a formulation by Ante Aikio that Proto-Mari *o appears before *l if the word does not begin with a glide.
  • Finnish pakkanen ‘frost’ ~ MariE W pokšə̑m ‘frost’
  • Russian леньгас ‘loafer’, Estonian lõngus ‘lout’ ~ MariE laŋga ‘lazy’. Paul Ariste connected the first two in a 1966 paper, even mentioning some Finnish words, but he didn’t mention the Mari at all even though it’s there staring one in the face.
  • MariE W paŋga ‘lump’ ~ Udmurt pog id. One would have thought the Mari would be included in the UEW (404) under *puŋkaKnollen, Beuele, Unebenheit’, as similar forms from across Uralic are listed there. MariE paŋga is in Paasonen’s dictionary, so it’s not like earlier researchers could have been unaware of this word. Even if one would prefer to see the Mari as a loan on account of its first-syllable *a, it’s curious that Bereczki didn’t include it in his 1992 list of Permian or Udmurt loanwords in Mari.

Mari /ŋ/ represented by Cyrillic <н>

In attestations of the Mari language from the 18th-century, Mari /ŋ/ tends to be represented with the Cyrillic letter <н>. Lots of manuscripts represent MariE jeŋ ‘person’ as <ен>, for instance. For more examples, see Alhoniemi’s 1979 commentary on the Mari wordlist of P. S. Pallas.

A colleague of mine found this odd, as he would have expected the sequence <нг>. Yet, denoting the sound [ŋ] in the same way as another single consonant has a long history. Consider Greek where the sequence [ŋg] is always spelled <-γγ->. Also, a samoyedologist once told me of a foreign colleague (Japanese, if I recall correctly) who kept hearing Nenets /ŋ/ as /g/; his ears simply couldn’t pick up on the nasal property of the consonant.

But if historically /ŋ/ has been confused by other peoples as either /n/ or /g/, the question remains why these Russian (and Russia-resident German) wordlist compilers constantly denoted Mari /ŋ/ with the symbol for /n/ and never for /g/. One reason for this may be that the compilers were already using Cyrillic <г> to represent Mari /ɣ/, which is a fricative, not a stop. Since the only other voiced velar sound in the language was a fricative, the velar stop /ŋ/ was heard as the closest stop to it: /n/.

But in the neighbouring Udmurt language, where the /g/ is a stop, not a fricative, 18th-century compilers still denoted /ŋ/ with the same symbol for /n/. D. G. Messerschmidt’s wordlist, which has been reprinted with a commentary by V. V. Napolskikh, has <Gurpuhn> for Udmurt dial. gurpuŋ ‘heron, stork’. (Note, however, how Messerschmidt denotes the sequence [ŋg] in <Ning-goron> for Udmurt dial. ńiŋgoron ‘woman’.)

So what else in these Uralic languages and in the native languages of these Russian and German compilers could have motivated the choice of the letter usually denoting /n/ and not the letter for /g/? Something worth thinking about.

Жгонский язык

While trawling back issues of the journal Sovetskoye Finno-Ugrovidenija for interesting reading on Mari, I came across a Russian dialect I had never heard of before, and which seems virtually unknown on the English-speaking web. As S. M. Strel’nikov writes in his 1978 article “Марийские элементы в жгонском языке” (Mari elements in zhgonsky jazyk):

Жгонским языком (от жгон ’шерстобит’) называют свой условный язык русские ремесленники Костромской области (пимокаты и портные), в недалеком прошлом занимавшиеся отхожим промыслом во многих губерниях России. Хотя численность носителей жгонского языка сокращается, его и сейчас помнят лица пожилого возраста во многих насееленных пунктах Нейского, Мантуровского, Макарьевского районов Костромской области, Варнавинского и Ветлужского районов Горьковской области.

Zhgonsky jazyk (from zhgon “woolspinner”) is the name by which Russian craftsmen in the Kostroma district (bootmakers and tailors) refer to their language; these craftsmen in the not-so-distant past were engaged in seasonal labor in many parts of Russia. Although the number of speakers of zhgonsky jazyk has declined, it is still remembered by elderly people in many settlements in the Ney, Manturov, and Makaryev regions of the Kostroma district, and in the Barnavin and Vetluga regions of the Gorsky district.

This language was an argot, meant to allow these craftsmen to communicate in secret when traveling about. Certainly the examples provided in this article are completely incomprehensible without glosses, e.g. Ши́до в плеха́нку пови́титься сохля́ть ‘I’ve got to head to the steam bath to wash’, Декни́ приты́лить ‘Give me a smoke’.

While zhgonsky jazyk drew on other languages such as Udmurt, German, Greek and Turkish, the Mari stock is prominent and Strel’nikov suggests that this argot arose on the basis of interaction between Russians and speakers of Northwestern Mari. Some zhgonsky jazyk words of Mari origin concern the numbers (e.g. ны́лик ‘4’ < MariNW nəl, канда́йша ‘8’ < MariNW kändäŋš) and weather (уре́ж ‘rain’ < MariNW jur, ю́кша ‘cold, winter’ < MariNW jükšem). Strel’nikov identifies altogether 44 items as derived from Mari, and some of them have gone amusing shifts in meaning as is common in these sorts of secret languages.

The two versions of Sergeev’s monograph on Mari manuscripts

O. A. Sergeev has dedicated much of his career to examining 18th and 19th-century manuscript word lists of the Mari language. In 2000 the Mari state press published his Mari-language overview of these treasures under the title Тошто марий мутер-влак.Cover of O. A. Sergeev’s 2000 publication Тошто марий мутер-влак In 2002, from the same publisher, his Russian-language monograph entitled Истоки марийской письменности appeared.

At a brief glance, one may be inclined to view the 2002 publication as simply a Russian translation of the original Mari-language work. Indeed, the table of contents is virtually identical; the two works discuss the same manuscripts in the same order, and then contain the same overall analysis of the materials from various perspectives. However, as I discovered in the course of my own research on Pallas’s Mari word list while initially using only the 2000 publication, there are differences between the two versions that mean that anyone looking into the history of Mari ought to go through both of them.

As an example, let’s consider Sergeev’s description of the manuscript Эрм. 577 №, инв. 802 held in the Russian National Library. The 2000 publication reads (pp. 35–36):

Ты очымо рукописят XVIII курым дене кылдалтын. Чылаже тудо 54 лаштык гыч шога. Рукопись марий-влакын илыме верла гыч Санкт-Петербургышто ямдылыме “Сравительный словарь всех языков и наречий” мутерлан колтымо материалжылан шотлалтеш. Каласаш кӱлеш: мутерысе шомак-влакым Российысе кугыжа Екатерина II шкеак возен. Кажне лаштык ик мут гыч шога, вара ты руш йылмысе шомакым “инородный” йылмылаш кусарыме. Марий йылмысе мут-влак 12-шо № дене пуалтыныт. Тылеч посна моло родо-тукым да пошкудо йылме-влакат улыт. Мутлан, 9-ше № дене эстон йылме кая, 10-шо № дене — удмурт, 11-ше № дене — чуваш йылме вераҥыныт.

Руш йылмысе мут-влак кӱшнӧ ончымо тыгаяк 286 реестре дене икгаяк улыт. Тылеч посна ик ойыртем уло: ты памятникыште руш йылме деч посна церковнославянский йылмын кышаже палдырна. Мултнан: вуй ’голова, глава’, шиндза ’глаз, око’, юкше ’стужа, холод, хлад’, молат. Руш йылмысе глагол, слово, речь шомак-влак марий йылмыш ик ’шомак’ мут дене кусаралтыныт.

17 мут деч моло шомак пелен ударенийым палемдыме.

Мут шагал умылан кӧра рукописьын могай наречий (але говор) негызеш ышталтмыжым каласаш неле.

The description in the 2002 publication (pp. 43–44) doesn’t mention that the Mari words are glossed in both Russian and Church Slavonic, and the presence of a particular Mari word for ‘cold’ is left out:

В данной рукописи хранятся материалы дла “Сравителного словаря всех языков и наречий”, написанные собственноручно Екатериной II. Итого — 54 л. Каждый лист содержит по одному заглавному слову, которое переведенно на “инородные” языки, напр., под № 12 даны марийские параллели, под № 9 — эстонские, под № 10 — удмуртские, под № 11 — чувашские и т. д. Реестр совпадает со словником 286 заглавных русских слов. Далее дается список 286 “черных” слов без перевода на “инородческие” языки.

Лексемы глагол, слово, речь переведены на марийский язык одним словом: шомакъ.

На лексемах, за исключением отдельных слов (17), поставлено ударение.

Скудный материал памятника не позволяет определить его диалектную основу.

For another manuscript, the 2002 publication mentions a misunderstanding between the Mari informant and the compiler of the word list with regard to the motion verb mijaš, which had not been discussed in the 2000 book. And in the general overview of the manuscripts, where Sergeev points out how some compilers offer a series of synonyms, the list in the Russian translation on page 69 cites fewer items than the 2000 original on page 69.

One might think these small differences, but one man’s small difference may be another’s vital piece of information.

The curious Mari word шнуй šnuj ‘holy’

The first time I ever came across Meadow Mari шнуй šnuj ‘holy’, I was quite struck by the word. It resembles nothing in Chuvash, Tatar or Russian, but the initial consonant cluster means that it cannot be a native Mari word.

While used in the modern literary language, the word is notably missing from the dialectal dictionaries compiled by late 19th-century or early 20th-century fieldworkers. As far as I can tell, the first dictionary to include the word is V. M. Vasil’ev’s Марий Мутэр (Moscow, 1926), where it is glossed ‘святой’ with the example sentence Шнуй пийал дэнэ йумо ий мучашы̆м шуйа маны̆т. The word is not present in the 1956 Mari–Russian dictionary, possibly on account of the anti-religious edicts of the time, but it is found in the 1991 Mari–Russian dictionary and the 10-volume Словарь марийского языка, where it seems to have long become an established feature of the literary language.

An initial cluster is also found in Eastern Mari spaj ‘beautiful, graceful’, a loan from Tatar zipa, in turn from Persian zibā (see Paasonen’s Ost-Tscheremissisches Wörterbuch pp. 112–113), so I looked for some Turkic or Perso-Arabic word of a similar shape for šnuj, but I eventually gave up. As the word has generally been overlooked in discussions of Mari etymology, there the matter rested for some time.

However, I recently saw that Raija Bartens took a look at šnuj in her paper “Marilaista Isä meidän–rukouksen käännöksistä”, a contribution to the Festschrift for Seppo Suhonen Oekeeta asijoo (Helsinki: Suomalais-Ugrilainen Seura, 1998). In coming across šnuj in her survey of Mari translations of the Lord’s Prayer through the ages, Bartens was just as baffled by the word was I was. Noting that it is used only in Christian contexts, Bartens asks if it isn’t simply a corruption of Russian священный.

Oddly, in his История марийского литературного языка (p. 137), I. G. Ivanov lists this word among terminology from the Mari pagan religion that fell out of active use during the rise of the Mari literary language. Not only was this word not used for Mari paganism, but it is the rise of the literary language in the 20th century that propelled this word from some extremely obscure origin to common use by Meadow Mari-speaking writers everywhere.

Mari words in Cheung’s Studies in the Historical Development of the Ossetic Vocalism

J. L. Cheung’s Studies in the Historical Development of the Ossetic Vocalism (Wiesbaden, Reichert Verlag, 2002), which goes well beyond what its title suggests, is in many respects an updating or refinement of Abaev’s Ossetian etymological dictionary. Cheung’s monograph also has an index for each of the languages, Iranian or otherwise, drawn on in the work. Unlike Abaev’s enormous, and mostly wrong, use of Mari, Cheung limits his etymologies to just four Mari words: βerɣe ‘kidney’, kutkə̑ž ‘eagle’, ož(o) ‘stallion’ and pire ‘wolf’.

Thus we are on much firmer ground than in Abaev’s dictionary, although Cheung again misrepresents the Mari word for ‘wolf’ as pirägy, and that is probably a borrowing from Tatar anyway.

Battle of the etymologists

The verb MariE püč́kampəčkäm ‘cut off’ is funny. In the Uralisches etymologisches Wörterbuch (367) the word is derived from a supposed Proto-Uralic *pečkä‑ (päčkä‑) ‘to cut’ on the basis of North Saami bæsˈkedi‑ ‘cut hair or wool off’ and Mordvin E M pečke ‘cut off, chop off’. Bereczki upholds this etymology in his Etymologisches Wörterbuch des Tscheremissischen (Mari) without mentioning any alternatives.

On the other hand, Fedotov in his Этимологический словарь чувашского языка (I 409) etymologizes Chuvash păčkă ‘saw’ on the basis of Turkic – namely the widespread *pïčak/bičäk ‘knife’ – and claims (again without mentioning any alternative) that MariW pəčkäm is a borrowing from Chuvash. Who is right here?

There is only one Uralic etymology in Bereczki where *pe‑ gives MariE pü‑, namely püńč́ö ‘pine’ < *penčä (UEW 727). Otherwise pü- in Mari is normally from *pä‑, e.g. pükš ‘hazelnut’ < *päškз (UEW 726–7). However, if we assume that the Proto-Uralic form was *päčkä, that would conflict with the Mordvinic forms, as Moksha Mordvin usually preserves PU *ä and does not raise it to e. I suppose that is why the UEW placed a question mark before the Mordvinic forms.

Can derivational morphology settle the question? The frequentative of this verb is püč́keẟem, and a quick search of the Mari–English Dictionary shows that ‑eẟem is overwhelmingly found in inherited Uralic vocabulary (or at least pre-Chuvash borrowings), not Turkic loanwords. It is not exclusively so – note joɣeẟem ‘flow’ < Chuvash and tojeẟem ‘hide’ < Tatar – but I would think it probable that MariE püč́kem is inherited.

Ultimately, however, with the resemblance between the Proto-Turkic and Proto-Uralic forms, we might have to take the dreaded notion of “sound symbolism” into account here, something which usually makes me want to drop the question entirely, leaving it for someone else with a greater gift for linguistics.

A curious source of /a/ in both Meadow and Hill Mari

When I first encountered the Mari words MariE paŋga, paklaka ‘stick, lump’, MariE W paŋgə̑ra ‘hard, stiff’, I was inclined to view them as loanwords due to the vocalism and a hunch that ‑ə̑ra and ‑laka are suffixes from some unknown substrate. (My ground for the latter suffix was a very un-Finno-Ugrian vacillation between l and r, cf. kakraka ‘hard, dry soil’ < kaŋga ‘poor, barren soil’.)

However, these words appear to be internally derivable in Mari. While Tscheremissisches Wörterbuch lists paŋga and paŋgə̑ra (paklaka is not in TschWb, only the Mari–English Dictionary and Paasonen’s Eastern Mari dictionary) under their own entries with no further comment, dozens of pages later under the entry for pu ‘wood’, one finds the verbal derivation puaŋeš ‘harden, become like wood’.

Surely we are dealing with the same word here. That the paŋga words were originally compounds, the first element of which was pu ‘wood’, would do much to explain the semantic range of paŋga, which includes ‘lump; stick; piece, figure (in a number of games), stick game; bead (on abacus), fishing float’. The vocalism is strange, however. Are there any other examples of the sequence *‑ua‑ producing /a/ in both Meadow and Hill Mari? (One would be tempted to explain other MariE a ~ W a matches with such an original sequence; could the verb manam ‘to say’ and the noun mut ‘word’ be thus related?)

And while we’re on the subject, MariE W peŋgə̑de ‘hard’ would be much more satisfyingly explained as related to these words and a compound with pu ‘wood’, than the traditional derivation from a supposed PU *piŋka ‘taught, tight’ (UEW 731, upheld in Bereczki’s etym. dictionary) where the semantics don’t seem to match.

Before I discovered the Mari words where pu ‘wood’ is visible, my candidate for the source of the Mari paŋga words was Iranian *fanai̯aka-, cf. Ossetian fäjnäg ‘board’. I suppose that link must now be abandoned, although the distribution of the Ossetian word – it is found elsewhere in Iranian only in Khwarezmian as fnyk ‘shield’, is certainly interesting.

Mari words in Abaev’s etymological dictionary of Ossetian

V. I. Abaev’s Историко-этимологический словарь осетинского языка (published in four volumes in 1958–1989) is quite famous and I was happy to discover a PDF on everyone’s favourite filesharing community for linguistics books. You can also order a paper version from some Russian online bookstores as print-on-demand. However, it wasn’t until I browsed the Helsinki library shelves that I discovered there was an index for it as well. The Указатель volume was published in Moscow in 1995.

(Furthermore, Abaev also published 22 pages of addenda and corrections to the dictionary as his contribution to the Festschrift for Ladislav Zgusta Historical, Indo-European and Lexicographical Studies ed. Hans H. Hock, Berlin: Mouton de Gruyter, 1997.)

The index contains sections for all the various languages Abaev dealt with, including individual Finno-Ugrian languages. As I am very interested in late East Iranian loanwords in Mari, I looked at what Mari words Abaev had mentioned. Below I present a list, with Abaev’s representation of the Mari (a jumble of transcriptions and dialect forms) replaced by the Tscheremissisches Wörterbuch headwords. Unfortunately, most of these can be treated as Chuvash or Tatar loanwords, inherited Uralic vocabulary or coincidential resemblances, and certainly not as the result of direct Iranian–Mari contact. Clearly the field has moved on since Abaev’s heyday.

Mari Ossetian Page Better etymology
alaša ‘gelding’ alasa id. I 44 < Tatar
čəgət cyxt ‘cheese’ I 328 Not in TschWb, but if Mari it would be < Chuvash
kə̑ńe ‘hemp’ kättag ‘cloth’ I 590
keńe gän id., kättag ‘cloth’ I 513, I 590
kerde ‘sword’ kard id. I 571
kož ‘spruce’ k’ozä ‘conifer shoot’ I 638 < PU *kose
kukšo xysk’ id. IV 270
mör ‘berry’ myrtkä id. II 141 < PU *mïrja
naməs namys id. II 155 < Tatar
pire ‘wolf’ biräğ id. I 263 < Tatar
pursa ‘pea’ pysyra ‘nettle’ II 248 < Chuvash
rüzem ‘to shake (trans.)’ rizyn ‘to shake (intrans.)’ II 418
rə̑βə̑ž ‘fox’ ruvas id. II 434
sokə̑r ‘blind’ soqqyr id. III 138 < Tatar
šu ‘bristle, fishbone’ syg ‘barb’ III 186
šüĺö ‘oats’ syl ‘rye’ III 194 < Chuvash
šur ‘horn’ sy id. III 181 < Proto-Iranian
toβar ‘axe’ färät id. I 451
tomaša ‘strange thing; commotion’ tamaša id. III 228 < Chuvash or Tatar
tul ‘stormwind’ tyfyl ‘whirlwind’ III 328 < Cv. tăvăl or Tat. tawïl
tumna ‘owl’ tojmon id. III 298 < Chuvash
tə̑rke ‘young pine’ tägär ‘maple’ III 252 TschWb says < Tat./FU?
umla ‘hops’ xymlläg id. IV 262 < Chuvash
uža ‘sells’ wäj id. IV 67 < PU *wosa, borrowed from PIE
βaraš ‘hawk’ wari ‘falcon’ IV 50
βürɣeńe ‘copper’ ärxy id. I 186
[eŋer-]βaze ‘fishing rod’ wis ‘rod, pole’ IV 111
βerɣe ‘kidney’ wyrg IV 123

It’s worth mentioning that Abaev’s supposed Mari word for ‘wolf’ is pirägy, clearly from MariE pire but in itself clearly erroneous. Abaev’s ghost word was later perpetuated in J. L. Cheung’s Studies in the Historical Development of the Ossetic Vocalism, p. 173, about which more later.

Have the Mari always been a forest people?

The self-image of the Mari during the last couple of centuries, reinforced by state-sponsored representations of Mari culture in the Soviet era, is that the Mari people have since time immemorial been dwellers of the forests. Marij kalə̑k – č́oẟə̑ra kalə̑k ‘the Mari people are a forest people’ is a saying that foreigners interested in the Mari come across early on.

I was struck today by how D. E. Kazantsev even used this image of the Mari as evidence against a proposed etymology for the ethnonym черемис, which the Russians and surrounding Turkic peoples used to denote the Mari but which was not used by the Mari themselves. In Формирование диалектов марийского языка (Йошкар-Ола: Марийское книжное издательство, 1985), p. 7–8, Kazantsev dismisses Gordeev’s derivation of the name from Iranian *čar ‘wander, roam’, thus ‘the [from the Iranian tribes’ view] people wandering nearby’, writing […] вызывает сомнение, что кочевой или бродячий образ жизни был мотивирующим признаком в этническом названий марийцев, всегда связанных с зоной лесов.

But are the Mari people truly a forest people? In his From Fugitive Peasants to Diaspora: the Eastern Mari in Tsarist and Federal Russia, Seppo Lallukka notes that the Mari often ended up dwelling deep in forests because they had been pushed there by the advance of the Russians across the Middle Volga region. The earliest significant depiction of Mari life, Nicolaas Witsen’s Noord en Oost Tatarye (in the second edition of 1705) indicates that many Mari lived in forests at that time, but he also mentions plains­dwellers and the breeding of cattle (and cattle appear to have been known to the Mari at least as early as the 1st millennium AD on the basis of the words skal ‘cow’ and šör ‘milk’). I don’t necessarily want to defend Gordeev’s etymology of черемис, but to claim that the Mari (who were at some point, after all, hunter-gathers) could not have been a mobile population and were perennially settled in forests is, I think, untenable.