The tangled etymology of tvorog

Chuvash turăx ‘sour milk’, according to Fedotov’s etymological dictionary, has a secure Turkic etymology, cf. Old Turkic tar ‘buttermilk’. This word in Chuvash or some other early Turkic language is probably the source of Common Slavonic *tvarogŭ (which was even further borrowed into German as Quark) as well as Hungarian túró. It is also claimed by Fedotov to be the origin of MariE torə̑k, MariW tarə̑k.

There is, however, another very similar Chuvash word both in meaning and in phonology that Fedotov does not connect with turăx, namely tăvara (Viryal tora) ‘small cheese’. (This is the source of MariE tuara, MariW tara ‘curd cake, cheese pastry, curd pancake’, Tatar tura ‘homemade cheese’.)

Looking at these two words together, I wondered if Cv. tăvara is the inherited Turkic word, while turăx is a reborrowing from Russian. Loss of a final voiced velar is a normal occurrence in Chuvash, cf. ura ‘leg’ < *aẟaɣ < *aẟak, so tăvara is to be expected.

While any connection goes unmentioned by Fedotov, both Chuvash words are treated together in Róna-Tas & Berta’s West Old Turkic under Hu. túró. In the course of their discussion, they write:

The WOT form toraɣ may be a denominal derivation from tor(ï) with the denominal suffix +rAk, which originally served as an intensifier and later a marker of the comparative degree. The voicing of the final -k is problematic here, because the suffix exists in Chuvash and is +rAx. It is, however, possible that the -x is a second intensifier in Chuvash, +rAk > +rAg > +rA > +rA+x.

Assuming reborrowing from Russian, could the unvoiced final velar in Chuvash turăx reflect the devoicing of final consonants in Russian after the loss of the yers? How old is that feature anyway?

The vocalism of the Turkic roots is problematic. Cv. tăvara assumes original first-syllable *-o/-u, but the word has been compared to Old Turkic tar with its original *a.

The ancient Indo-European comparanda τυρός ‘cheese’ and Avestan tūiri ‘cheeselike milk, whey’ (see Beekes Etymological Dictionary of Greek) make me wonder if this is a steppe Wanderwort.

Ablaut-motivated exceptions to Cowgill’s law

A decade ago I set about making flashcards for Indo-European sound laws and was always fond of Cowgill’s law for its extremely succinct wording: [in Greek] /o/ becomes /u/ between a labial and a resonant.

There are plenty of examples of this sound change listed in introductions to comparative IE linguistics or the history of Greek (νύξ ‘night’ < *nokʷts etc.), but it never hit me that this law did not apply to perhaps the largest source of Greek /o/: an Indo-European /o/ ablauting with /e/ and the zero grade.

This eureka moment came when I paused to consider Homeric Greek βουκόλος ‘cowherd’, which should go back to PIE *gʷowkʷolh₁os (this is the etymology given by Beekes in his Etymological Dictionary of Greek). The second labiovelar does lose its labial component as one would expect in a Cowgill’s law environment, but the following /o/ vowel is not raised to /u/.

This particular word is deceptive, however. If we look at the Mycenaen evidence, we find in Mycenaen qo-u-ko-ro that the second-syllable labiovelar of this word had already lost its labial component through dissimilation from the preceding /w/ and became a plain velar.

In most instances, however, the labiovelar in Greek words based on *kʷolh₁os was preserved down to post-Mycenaean times, and yet Cowgill’s law still fails to apply. Consider πολος ‘axis, pole, vault of heaven’ < *kʷolh₁os, τρίπολος ‘thrice-ploughed’ < *tri-kʷolh₁os, νυκτίπολος ‘traveling by night’ < *nokʷti-kʷolh₁os and, if one wants to open the can of worms that is the Greek word for ‘horse’, ἱππόπολος ‘horse-taming’ < Proto-Greek *hikkʷo-kʷol(h₁)os.

PIE *kʷolh₁os in these words is a deverbal, o-grade noun. The normal-grade root *kʷelh₁- was inherited by Greek in the verb πέλομαι, which by the time of its attestation had shifted from the meaning ‘go round’ to ‘to stir, take place, be’.

One finds a similar survival of unraised /o/ in πόρος ‘ford’ < *per, φόρος ‘tribute’ < *bher and βορός ‘gluttonous’ < gʷerh₃. For these three roots, there is an array of other derivational forms showing different ablaut grades.

At the time Cowgill’s law operated, the ablaut system of three grades must have been so productive that it blocked the raising of /o/ to /u/ in words Proto-Greek speakers felt to be related. As the majority of ablaut-derived /o/ were found in words that did not have a labial + resonant environment, there would be strong analogical pressure to avoid such a raising of ablaut-derived /o/ in the small stock of roots that did.

Various Turkic–Mongolic etymological observations

Preparing to study Mongolian from Krueger’s An Introduction to Classical (Literary) Mongolian (Wiesbaden: Harrassowitz, 3rd edition 1993), I’ve been re-reading the Routledge Language Family Surveys volume The Mongolic Languages ed. Juha Janhunen. Below are some musings on and follow-ups to trivia within.

Examples of some crucial [Khalka] consonant contrasts: ad [at] ‘demon’ vs. at [aʰt] ‘castrated camel’; dal [taɮ] ‘seventy’ vs. tal [tʰaɮ] ‘steppe’.

So modern Mongolian is one of those languages that, instead of a voiced–unvoiced distinction in dentals that I could actually pronounce, has an aspirated–unaspirated distinction that I’ll never get down. That’s a damn shame.

[Turkic borrowings in Mongolic] often show a specialized meaning, whereas the native [Mongolic] words have a more general semantic profile, cf. e.g. Mongolic *xüsün ‘hair’ vs. * ‘hair of a horse’ ← Bulgharic kïlka = Common Turkic *kïl (qïl) ‘hair’.

The ordinary Chuvash word for ‘hair’ today is ҫӳҫ. However, for Russian конский волос ‘horsehair’, the Skvortsovs’ dictionary gives лаша хӗлӗхӗ. For Cv. хӗлӗх, Fedotov’s Этимологический словарь чувашского языка gives a wide array of Turkic cognates, but they are all glossed as ‘horsehair’, so it’s unclear to me on what grounds Claus Schönig in the passage I’ve quoted believes it ever meant ‘hair’ in general.

In the Common Turkic branch, rhotacism, lambdacism is generally absent, but it is occasionally observed in preconsonantal position, which makes the dating of certain loanwords problematic, cf. e.g. Mongolic *buxas ‘pregnant’ (from Common Turkic *bugaz id.) vs.‑ ‘to cut the throat’ (from either Bulgharic or Common Turkic, cf. Common Turkic *bogaz ‘throat’).

That Bulgar Turkic had a cognate word for ‘throat’ showing rhotacism is attested by Chuvash пыр id.

Mongolic ulus ← Common Turkic uluš (later replaced in most Turkic languages by a reborrowing from Mongolic).

There is an informative entry on Common Turkic *uluš/ulus on page 152 of Clauson’s A Dictionary of Pre-Thirteenth Century Turkish, which notes that the original Turkic form uluš seems to survive only in Karaim.

Mongolic *kerbish ‘brick’ ← Common Turkic *kärpič

The Common Turkic is the source of Russian кирпич. It must say something of the material poverty and fondess for wooden buildings of the Russians of old, that they had to take the word for ‘brick’ from a population generally associated with yurts.

The early Kipchak source Codex Cumanicus exhibits [Mongolic] borrowings like abaɣa ‘uncle’, čïray ‘face’, ebäk ~ elpäk ‘very much’, yada‑ ‘to get tired’, qurulta ‘assembly, council’, manglay ‘forehead’, nögär ‘follower’, and qaburqa ‘rib’.

For what it’s worth, several of these are commonplace in Tatar as well, namely абый, чырай, бик, маңгай and кабырга.

Mongolic *köper > *köxer ‘proud’ > ‘happy’ vs. Turkic *küpez (> *kübez) ‘proud’, Mongolic *köperge > *köxerge ‘bridge’ vs. Turkic *köprüg (*köbrüg).

Of the first set of words here, I’m tempted to claim some connection to Tatar чибәр ‘beautiful’, with cognates in languages of the Volga region meaning ‘happy’. Could the k‑ of the Mongolic or Bulgar word cited above have shifted to an affricate before a front vowel in some other language that was the source of the Tatar? However, I don’t seem to own any etymological reference that describes this possibility. Äxmat’janov’s Татар теленең кыскача тарихи-этимологик сүзлеге suggests only that the Tatar is borrowed from a Mongolic cegeber ‘white, clean’.

For the second set of words, I’ve long suspected a connection to Greek γέφῡρα, but the entry in Clauson on page 690 mentions no connection between the Turkic and other language families (except the loan in Mongolic), mentioning only morphologically Dev. N. fr. köpür‑ [‘to froth, to foam’] but with no obvious semantic connection. On Greek γέφῡρα, Beekes on page 269 of his Etymological Dictionary of Greek suggests the Greek is borrowed from Hattic hammuruwa ‘beam’, with all instances of the words in Homeric Greek representing ‘beam’ and the meaning ‘bridge’ is attested only later. However, if a meaning ‘bridge’ is attested for this word by the mid 1st millennium BC, would that not give plenty of time for it to be borrowed into an unknown Iranian language of Central Asia and then picked up by Turkic?

The lesser-known W. Sidney Allen

Any student of classical languages with a linguistics bent will delight at discovering W. Sidney Allen’s books Vox Latina and Vox Graeca that reconstruct the pronunciation of Classical Latin and Greek, respectively. Cambridge University Press has published them in relatively cheap paperbacks. However, there are two more works by this scholar that that don’t get anywhere near the attention they deserve, even though they are logical next steps.

The first is Accent and Rhythm: Prosodic Features of Latin and Greek (Cambridge University Press, 1973). Here W. Sidney Allen takes the linguistic reconstruction of Greek and Latin one step further from Vox Latina and Vox Graeca to encompass suprasegmental aspects of these languages. This book does demand a greater understanding of theory (whereas the earlier books expected little more than some knowledge of IPA), and it takes some work to apply Allen’s insights to one’s own enunciation.

The second book treats what is historicaly the third important classical language for Indo-European studies, Sanskrit. Allen’s Phonetics in Ancient India (Oxford University Press, 1953) was published years before Vox Latina and Vox Graeca, and is organized somewhat differently in that it is mainly a retelling of the already very detailed ancient Indian sources for Sanskrit pronunciation. However, Allen does engage in some detective work to clarify matters obscure in the ancient grammarians, such as the pronunciation of the visarga.

Xenophon comix

In one of the odder installments in a university press series, volume 16 of Odense University Classical Studies is a graphic novel adaption of Book I of Xenophon’s Anabasis, where the original Greek text is paired with illustrations by Minna Winsløw. Were this somewhat larger (it is only 25 pages long, heavily abridging the text) and if the Greek were written with better calligraphy, I could see this motivating at least some students out there.

The first page from the Odense University Press graphic novel of Xenophon's Anabasis"

You can find this in a university library near you – or probably not – under the title ΑΝΑΒΑΣΙΣ (Odense Universitetsforlag, 1991) ISBN 8774928007.

A Uralic loanword in late Proto-Indo-European?

I may have come across such etymologies before, but as far as I remember, this is the first proposal I’ve seen of a Uralic loanword in Proto-Indo-European. In Ananta Śāstram: Indological and Linguistic Studies in Honour of Bertil Tikkanen ed. Klaus Karttunen (Helsinki: Finnish Oriental Society, 2010), Asko Parpola has this to say on the etymology of Finnish kaivaa ‘dig’:

The Finnish words kaiva-a ‘to dig’ and kaivo ‘digging, well, pit’ have cognates in Finnic languages, in Saami and the Volgaic and Permic languages. Ante Aikio has shown that Proto-Finno-Ugric *kajwa- can be regularly connected with Proto-Samoyedic käjwa ‘spade’, as the change *a > took place in Samoyedic before a tautosyllabic palatal consonant, thereby settling an old problem, the history and material of which is fully discussed by Aikio. Hence the etymon is an archaic Uralic nomen verbum.

What I offer here is not a new etymology, but simply a reference to an old etymology proposed as early as 1920 that was not included in the indexes of etymologically treated Finnish words by Donner and Erämetsä, and so has escaped notice in SKES and SSA. K. F. Johansson had reconstructed an archaic Proto-Indo-European heteroclitic noun *kaiw-r̥-t (nom.) ~ *kaiwn̥n-eś (gen.) on the basis of Greek and Old Indo-Aryan. Hesychius records καίατα in the sense of ‘pits, excavations, trenches, ditches’ (ὀρύγματα) or ‘landslide chasms caused by earthquake’ (ἢ τὰ ὑπὸ σεισμῶν καταρραγέντα χωρία) The plural καίατα is supposed to stand for καίϝατα, from the singular καίϝαρ. Old Indo-Aryan kevaṭa- ‘pit’ is attested in a single occurence in the oldest text, Rigveda, 6,45,7; Old Indo-Aryan e goes back to Proto-Aryan *ai and *rt has often become retroflex *ṭ. Pokorny accepts the comparison and reconstructs for Proto-Indo-European *kaiwr̥t *kaiwn̥-t. Thomas Burrow and Manfred Mayrhofer have considered the scanty evidence in both Old Indo-Aryan and Greek as too uncertain for the assumption of a PIE hetercliton. Still, Mayrhofer thinks it is possible that the words are related. Herbert Petersson also emphasizes that no trace of this etymon is found in other Indo-European languages — and Frisk points out that no corresponding PIE verbal root can be traced — while the root structure too, with a diphthong following by -w-, also looks peculiar for PIE. Petersson therefore takes this to be one of the rare cases where Proto-Indo-European is likely to have borrowed from Proto-Finno-Ugric. Mayrhofer refers to Petersson’s suggestion as noteworthy but unconfirmed. However, the confirmed Uralic origin of kajwa- and the archaic appearance of the word on both sides gives new significane to Petersson’s hypothesis.

(The title of Parpola’s contribution to this volume is ‘New Etymologies for Some Finnish Words’, pp. 305–318. In quoting it here, I have slightly abridged the text and left out the parenthetical citations for the sake of readabiity.)

Gospel of John in Homeric Greek

During the evening service on Easter (the so-called Agape Vespers), the Orthodox Church has a tradition of reading the gospel passage for that day (John 20:19–25) in many different languages. In Greece, one of the versions sung is a rendering by St Nicodemus the Hagiorite (1749–1809) which casts the passage in the Homeric Greek and dactylic hexameter of the Iliad and Odyssey.

Ὄφρακε νωιτέροισιν ἐν οὔασι πάγχυ βάλωμεν
θέσφατον, ἱμερόεσσαν, ἁγνὴν Εὐάγγελον ὄππα
μειλίξωμεν Ἄνακτα Θεὸν μέγαν, οὐρανίωνα.
Ἰθυγενεῖς. Σοφίη. Εὐαγγελίοιο κλύωμεν.

Εἰρήνη χαρίεσσ’ ἐπ’ ἀπείρονα δῆμον ἐσεῖται.

Ἐκ δ’ ἄρ’ Ἰωάννοιο τόδ’ ἔστι βροντογόνοιο.

Ἄλλ’ ἄγετ’ ἀτρεμέσι χρησμοὺς λεύσωμεν ὀπωπαῖς.

Εὖτε δὴ ἠέλιος φαέθων ἐπὶ ἕσπερον ἦλθε
καὶ σκιόωντο ἀγυιαὶ ἐπὶ χθονὶ πουλυβοτείρῃ,
ἥματι ἐν πρώτῳ, ὅτε τύμβου ἆλτο Σαωτήρ,
κλῃισταὶ δὲ ἔσαν θυρίδες πυκινῶς ἀραρεῖαι,
βλῆντο δὲ πάντες ὀχῆες ἐυσταθέος μεγάροιο,
ἔνθα Μαθηταὶ ὁμοῦ τε ἀολλέες ἠγερέθοντο
μυρόμενοι θανάτῳ ἐπ’ ἀεικέι Χριστοῦ Ἄνακτος
καὶ χόλον ἀφραίνοντα Ἰουδαίων τρομέοντες,
ἤλυθε δὴ τότε Χριστὸς Ἄναξ θεοειδέι μορφῇ,
ἔστη δ’ ἐν μεσάτῳ ἀναφανδὸν καὶ φάτο μῦθον·
Εἰρήνη ὑμῖν φίλη, ἡσυχίη τ’ ἐρατεινή.
Ὡς εἰπὼν ἐπέδειξεν ἑὴν πλευρὴν ἠδὲ χεῖρας.
Γήθησαν δὲ Μαθηταὶ ἐπεὶ ἴδον Εὐρυμέδοντα.

Τοὺς δ’ αὖτις προσέειπεν Ἰησοῦς οὐρανοφοίτης·
Εἰρήνη ὑμῖν φίλη, ἡσυχίη τ’ ἐρατεινή.
Ὡς ἐμὲ πέμψε Πατὴρ ὅς ὑπέρτατα δώματα ναίει,
ῳδ’ ἐγὼ ὑμέας ἐς χθόνα πέμπω εὐρυόδειαν.
Ὡς ἄρα φωνήσας Μύσταις ἔμπνευσ’ ἀγορεύων·
Πνεῦμα δέχνυσθ’ Ἅγιον, φαεσίμβροτον, ὑψιθόωκον·
Ὧν μὲν ἀτασθαλίας θνητῶν ἀφέητ’ ἐπὶ γαῖαν,
τοῖσι νύπου ἀφίενται ἐς οὐρανὸν ἀστερόεντα·
ὧν δ’ ἄρ’ ἐπεσβολίας ὑππερφιάλων κρατέητε,
τοῖσιν ἁλυκτοπέδῃς κεῖναι σθεναρῇς κρατέονται.

Θωμᾶς δ’ ῳ ἐπίκλησις ἅπασι Δίδυμος ἀκούειν
οὐχ ἅμα τοῖς ἄλλοις Μύσταις πρὶν ὁμώροφος ἔσκε
Ἰησοῦς ὅτ’ ἔβη εἴσω μελάθροιο ἑταίρων.
Ἴαχον οὖν ἄλλοι τούτῳ ἐρίηρες ἑταῖροι·
Εἴδομεν ὀφθαλμοῖσιν Ἰησοῦν παγκρατέοντα.
Τοὺς δ’ ἀπαμειβόμενος Θωμᾶς προσέφησεν ἀτειρής·
Ἴχνια ἤν μὴ ἴδω μετὰ χείρεσιν ἡλατορήτῃς,
δάκτυλον ἐμβάλλω τε ἐκείνου ἔνδοθι χειρός,
χεῖρα τ’ ἐμὴν εἴσω πλευρῆς οἷ ρεῖα βαλοίμην,
οὔποτε ὑμετέροισι λόγοις κεφαλῇ κατανεύσω.

Like the Latin translation of the Kalevala by Tuomo Pekkanen, this could only have been made by someone with a great love of epic and a lot of time on their hands.

Linguistics and classical teaching

The Winter 2007 issue of The Classical World featured a collection of papers under the heading ‘The Linguistic Edge: Using Linguistics to Enrich the Teaching of the Classics’:

  • Joshua T. Katz, ‘What Linguists are Good For’
  • Egbert J. Bakker, ‘Time, Tense and Thucydides’
  • Mary R. Bachvarova, ‘Actions and Attitudes: Understanding Greek (and Latin) Verbal Paradigms’
  • Rex Wallace, ‘Using Morphophonology in Elementary Ancient Greek’
  • Robert J. Littman, ‘Linguistics and the Teaching of Classical History and Culture’
  • Gregory Nagy, ‘The Fire Ritual of the Iguvine Tables: Facing a Central Problem in the Study of Ritual Language’

These have some useful insights for those who are learning (or, in my case, re-learning) Greek with a background in linguistics. After reference to Spanish’s use of the se pronoun in verbs, Bachvarova’s paper gives a list that I wish I had as an undergraduate:

One can find the corresponding activities of all these verbs of course, but the fact is that this active form, the form one looks up in the dictionary, is usually much more rare than the middle form for actions which can be conceived of as spontaneous events, which causes my students, who are used to thinking of the active voice as the ‘basic’ voice, no end of trouble in recognizing and finding verbs in the dictionary, as with naturally aoristic event-types. This has therefore become one of my key points in introducing the middle: certain situation-types are more or less naturally middle, and even if you are forced by grammar books and dictionaries to learn its active form, you should not expect to see it very often, and sometimes there is no active form at all; these are deponent verbs.

(13) Typically Middle Situation-Types:
Grooming/body care ἀλείφομαι
Nontranslational motion τρέπομαι
Change in body posture καθίζομαι
Translational motion οἴχομαι
Naturally reciprocal events διαλέγομαι, μάχομαι
Indirect middle κτάομαι
Emotion middle φοβοῦμαι
Emotive speech actions ὀλοφύρομαι
Cognition middle οἴομαι, πυνθάνομαι
Spontaneous events φύομαι, τήκομαι

Greek and Latin verse composition

For the past several weeks I’ve been working with North and Hilliard’s classic Greek Prose Composition to brush up on my Greek. It has only now occurred to me that if all those composition workbooks I encountered as an undergraduate thought to specify prose composition, then there must have been exercises of writing verse in Greek and Latin as well. Indeed, an old post at Textkit has an ample list of verse composition books published during the golden age of English classical studies.

The textbooks of this era are now in the public domain, so they might be available from various digitalization projects. For example, Gepp’s Progressive Exercises in Latin Elegiac Verse of 1880 can be read at Google Books.

These might provide hours of entertainment for those who can’t get enough composition exercises. I however, being something of a modernist snob who doesn’t believe poetry even really existed before the late 19th century, would not find it enjoyable to write within all those musty old metres.

An unexpected Greek typeface

In the course of updating the installation of the LaTeX typesetting system on my computer, I was randomly reading the supplied documentation and discovered a remarkable Greek typeface that I had never come across before. This typeface is most notable for its use in printed editions of the Philokalia, the great compendium of Orthodox teaching compiled between the 4th and 15th centuries. Here’s a sample of a passage from Aristotle’s Poetics from the documentation:

A Greek passage set in the Philokalia typeface

This typeface places accents after capital letters. Furthermore, it contains over 40 ligatures, many of which you wouldn’t be able to guess.

Ligatures in the Philokalia typeface

To those familiar with Greek manuscript writing and the introduction of printing to the Greek-speaking world, this is probably old-hat. But because my own studies of Greek have been limited to the classical canon typeset using standard typefaces like Porson and Lasergreek, I’ve never encounted this exotic writing system before, and a Greek original inspiration certainly helps explain the unnecessary flourishes of some printed Church Slavonic texts.

(For TeXnicians, the relevant package is Apostolos Syropoulos’ philokalia, supplied with XeTeX.)