Karamanlidika orthography

I blogged about Phanariot in the last post, but what I actually wanted to talk about was something far more tangential.

Phanariot, as we discussed, was filled to the brim with Turkish loanwords. Phanariot was still Greek, and it was still written in Greek script. That included the Turkish loanwords in the Greek.

But the Phanariots using Turkish words pronounced them in accurate Turkish; and they were concerned to write Turkish words in Greek script with phonetic accuracy. So they employed the conventions of Karamanlidika.

The Karamanlides were a Turkish-speaking Greek Orthodox people living in Karaman and Cappadocia; being Christians, they were subject to the 1924 population exchanges, and were resettled in Greece. Whether they were originally Karaman Turks, or Turkicised Cappadocian Greeks, or, much more plausibly, both, is hard to know, and not terribly relevant here anyway.

What is relevant is that script followed creed in the Ottoman world; so if a Greek Orthodox Christian spoke Turkish, and was literate, they would read Turkish not in the Arabic script, like their Muslim colinguals, but in Greek script. As an extension, Karamanlidika was the name given to Turkish written in Greek script — whether it was written by or for Karamanlides, or by the Greek-speaking Phanariots, who sprinkled their Greek so generously with Turkish.

With the opening up of Greek academe to the Ottoman past, there has been much research into Karamanlidika in recent years, and Evangelia Balta of the National Hellenic Research Foundation has been the main researcher active. Her website appears to be down currently, and the most accessible resource online for Karamanlidika is a somewhat unexpected source.

The Karamanlides moved to Greece. Their speaking of Turkish was frowned upon in the new country, and was not something they passed down. They did pass down other aspects of their culture, though, like their cuisine. One of those they passed their culinary heritage down to was Fanis Theodoropoulos, who has opened a Karamanli restaurant in downtown Athens: Τα Καραμανλίδικα του Φάνη, “Fanis’ Karamanli Foods”. And with the web address http://karamanlidika.gr, Fanis has felt it proper to include a blog on the restaurant website, covering not just the latest news and offerings of the restaurant, but also the culture and language of the Karamanlides and the Cappadocians. Including their script.

Thus, Samples of books printed in Karamanlidika, from the 16th to the 19th century. The 16th century is a reference to Martin Crusius printing a Karamanlidika text in his 1584 Turcograecia, the first Western study of Modern Greek. Other than that, the samples are almost all 19th century, with one text from the late 18th.

Now, there are several phonemes of Turkish that cause difficulty transcribed into Greek: none of <c ç h ı ş ö ü> /dʒ tʃ h ɯ ʃ ø y/ are present in Standard Greek. But as it turns out, <b d g> also pose a long-standing difficulty for transliteration in Modern Greek. Modern Greek has the phones, but not necessarily the phonemes: in many dialects of Greek, and in the Standard Modern Greek of older speakers, they occur only prenasalised, as reflexes of prenasalised stops. In other dialects, and in younger Standard Modern Greek, the prenasalisation drops off: <μπ ντ γκ> Ancient /mp nt ŋk/ > Older Standard Modern Greek /mb nd ŋɡ/ > Younger Standard Modern Greek /b d ɡ/.

Which means, sure, if you want to transcribe /b d ɡ/, you’ll use <μπ ντ γκ>. But given the history of Greek, and the variation in pronunciation, you’ll also use them to transcribe /mb nd ŋɡ/. And, indeed, /mp nt ŋk/.

The pronunciation of all of these has fallen together in Modern Greek, and Greek-speakers are not necessarily aware of it. Dante, for example, was transliterated as Δάντης, <ðantis>; but the Modern pronunciations of the name are /ðandis/ and /ðadis/. Greek loves using the German loanword lumpen to refer to “trash culture”, from the Marxist notion of the Lumpenproletariat, the underclass. But they’ve only ever seen it as λούμπεν, and they know less German than they do English, so they’ve pronounced it as they read it. Hence when time came for a Greek satirical website to adopt the word as its name, it inevitably ended up as https://luben.tv.

What this means is that if you’re trying to transcribe /b d ɡ/ in Greek, <μπ ντ γκ> might have ended up the only choice in contemporary Greek, especially given how integral /(m)b (n)d (ŋ)ɡ/ are to the language. But if you are not writing Greek at all, especially back in the 18th and 19th century, you don’t have as strong a motivation to lean on the Hellenic <μπ ντ γκ>. Especially when you’re already having to deal with /dʒ tʃ h ɯ ʃ/ anyway.

There is not a unified Karamanlidika transcription, but there are two main conventions one can see in the editions: one with diacritics, the other without. (The one diacritic it does admit is the diacritic 19th century Greek already admitted for Demotic: ι̮ for [j].) The transcription without diacritics fails to make some distinctions, like that between /mb/ and /b/ (both μπ), or between /ʃ/ and /s/ (both σ); it does however take advantage of the fact that Turkish has no /ð/, to transcribe /d, dʒ/ as δ, δζ. And at least by the 19th century, to judge from the samples, Karamanlidika had settled on using a spare Greek grapheme for /i/, η, to transcribe /ɯ/, in both the diacritic-based and the non-diacritic transliteration.

… Except when there’s an actual Greek word in Karamanlidika. When the word is Greek and not Turkish, Greek historical orthography is respected, and eta means /i/. So in fact, to read Karamanlidika accurately, you are sort of expected to know something about Greek anyway.

The rulebook of the Cappadocian Educational Fraternity, for example, has the following code switch (Greek words in boldface, and I am guessing at the Turkish via Google Translate):

• συνιστᾶται … Καππαδοκικὴ Ἐκπαιδευτικὴ Ἀδελφότης, γι̮άνι Καΐσεριε ἐπαρχίασηνην Οὐχουββέτι τεδρισιερί τεσ̇κὶλ ὀλουνμούσ̇δο̇υρ
• Kapaðokiki Ekpeðeftiki Aðelfotis, yani Kayseriye eparxiasının Uhuvveti tedrisiyeri teşkil olunmuşdur
• The Cappadocian Educational Fraternity, that is the educational brotherhood of the district of Kayseri, has been created

In a Greek word like Καππαδοκικὴ, eta is /i/; in a Turkish suffix like -σηνην, even if it is attached to the Greek word ἐπαρχία, eta is /ɯ/. Similarly, the list of Bible books—

—contrasts the Gospel of Luke, ΛΟΥΚΑΣΗΝ Lukasın, with the Gospel of John, ΙΩΑΝΝΗΣΙΝ Ioannisin. Eta is /ɯ/ in the Turkish suffix of Lukasın; it is /i/ in the Greek name Ioannis.

But I’m more interested in the diacritics used with Karamanlidika. Karamanlidika with diacritics adopted the overdot as a universal diacritic. So <χ̇ σ̇ π̇ τ̇ τ̇ζ ο̇υ ο̇> are /h ʃ b d dʒ y ø/. We have just seen δ used next to τεσ̇κὶλ: because it could not mean /ð/, it was still used to mean /d/ in some transcriptions, though not others: the Bible book list is titled ΑΧΤ̇Η ΑΤΙΚ ΙΛΕ ΑΧΤ̇Η Τ̇ΖΕΤ̇ΙΤ̇ΙΝ ΦΙΧΡΙΣΤΙ Ahdı Atik ile Ahdı Cedidin fi-Xristi “The Old Testament and the New Testament of Christ”. Similarly, while the Bible books use τ̇ζ for /dʒ/, they do not use τσ̇ for /tʃ/: Turkish has no /ts/ (hence it transliterates Tsipras as Çipras), so the Bible is content to use τζ (/ts ~ dz/ in the 19th century) for /tʃ/: Ο̇ΥΤΖȢ̇ΝΤ̇ΖȢ̇ ΠΑΤ̇ΙΣ̇ΑΧΛΑΡ <Ọutzọunṭzọu Paṭiṣaxlar> Üçüncü Padişahlar “3 Kings” (corresponding to Western 1 Kings). And there does not seem to be any use of γ̇: γγ/γκ have been accepted as /ɡ/, since Turkish has no /ŋɡ/.

The Karamanlidika overdots for /b d/ have made enough of an impression on Greek scholars, that they have been reused rather far from Greek transliterations of Turkish. The plays of the Cretan Renaissance were written in Roman script, using Italian orthographic conventions; Italian of course differentiates /b mp mb/, and just as Phanariot in Greek script preserved the non-Hellenic distinctions of Turkish loanwords, so too did the Italianised Cretan of the Renaissance preserve the non-Hellenic distinctions of Italian loanwords. But those texts still ended up published in Greek script, both at the time and in modern scholarly editions. And in at least one instance (I think it’s Alfred Vincent’s 1980 edition of Fortounatos, edited from the Roman script autograph manuscript), the Karamanlidika dots are used to differentiate the original’s <mp> and <mb>.

In the 18th century, though, there was a different way of writing /ʃ/. This merited a passing comment on p. 33 of Peter Mackridge’s paper on Phanariot:

For [b] [d] [ʃ] [dʒ] some writers use the diacritics that had been developed for use in karamanlidika earlier in the eighteenth century (π̇, δ̇, σ̈́, τ̇ζ), but most make do with the unadorned Greek alphabet: thus [kurdízo] is variously represented as κουρδίζω, κουρντίζω and κουρτίζω.

He adds in footnote: “For [ʃ] the sigma is in fact surmounted by three dots in a triangular pattern, but I am unable to reproduce this here.”

σ̈́?

I’ve asked Peter, and he’s sent me a sample from the satirical comedy Το σαγανάκι της τρέλας. (Not “The frying pan of madness”, let alone “The saganaki of madness”: contemporary Greek σαγανάκι “small frying pan”, and any dish prepared in a small frying pan, like fried cheese, is a diminutive of σαγάνι < Turkish sahan “copper dish”. The Turkish word here is the unrelated sağanak: “The storm of madness”.) The comedy is attributed to Rigas Feraios, and was published in Lia Brad Chisacof. 2001. Ρήγας. Ανέκδοτα κείμενα, Athens. the text is published alongside the manuscript, and he has sent me two instances of the novel diacritic in question: pp. 158–159 ϗʹ μεσ̈́αλάδες “& torches” (Turkish meşale), and pp. 176–177 μεσαλοσ̈́άνον (= Greek μεγαλουσιάνον “bigwig”; in that case, the character is being used for [sj] < [ʃ], presumably a dialectal pronunciation within Greek.)

This is what the two words look like in the manuscript:

Peter has told me that “What this scribe (who is no doubt the author) actually writes is not three dots but more like an acute accent flanked on each side by a dot.” Other instances may have been triple dot, but this instance is more like the diaeresis + acute that he resorted to in print. Maybe, though they look like triangular triple dots to me in the manuscript anyway.

And there’s a reason they should. I didn’t give the entire footnote above. He continues:

Christodoulos Christodoulou informs me that this use of three dots follows Arabic practice, šīn [ʃ] being distinguished from sīn [s] in the same way in Arabic script.

That diacritic is from Arabic: it is the distinction between س and ش. There is precedence for mixing diacritics and letters from different scripts; Samaritan for example has combined Samaritan letters (related to Hebrew) with Arabic vowel diacritics. This is the first time I’ve seen it in Greek.

And there’s good reason Peter Mackridge had difficulty rendering a three-dot sigma in his 2017 paper. Unicode has the three-dot Arabic diacritic, U+06DB ARABIC SMALL HIGH THREE DOTS. But combining Arabic diacritics with Greek or Roman script is disastrous: the diacritics are designed for a completely different letter height. Luckily, Unicode does offer a triple dot diacritic compatible with Roman (and Greek): U+1AB4 COMBINING TRIPLE DOT. Unluckily ,the character was added in Unicode in 2014, which means font support for it is still minimal: among the fonts I have installed (and I have a lot), the diacritic only turns up in Google’s Noto fonts and Dehuti, and sigma with triple dot only looks presentable in the latter: