The Trakatroukides and their language

This is a complicated story no matter how one tells it and from what vantage point, but I’ve put this off a few months too many. So, following on from the preceding post on the Trakatroukides, let’s take the story from the top. Again, thanks to Butcher of Yore for doing all the research.

After the Greek-Turkish war of 1919–1922 and population exchanges of 1924, thousands of refugees settled in Greece. These refugees were deemed Greek because they were Greek Orthodox, but they did not all speak a form of Greek. As I wrote at the Other Place, a good number of them spoke Turkish: over 100,000 first language speakers out of 1.3 million refugees—191,254 reporting Turkish as their language in the 1928 census (Google Books excerpt below), minus 84,669 indigenous Turks in Thrace (p. 17 of Featherstone & Papadimitriou’s presentation):

But language was not what the Ottomans or their subjects used to define identity.

The villages that were abandoned are now dwelled in by the Muslims that stayed, and one such village is Kızderbent, “Maidenpass”, in the region Greeks call Bithynia, 28 km away from Nicaea, which the Turks call İznik.

View Larger Map

The refugees from Kızderbent, who go by the name of Trakatroukides or Rokatzides, were scattered in Greece. There are two Facebook groups for refugees from the village (1, 2), and per this post on the second, most settled in Valtotopi, Kilkis prefecture; others ended up in the nearby villages of Ryzia, Gerakona, Fillyria, and Polypetro, but also further afield, in Nikomidino, Kallithea (Thessalonica prefecture), New Triglia (Chalkidike), and Roditis (Rodopi): this local newspaper article is on a get-together of the Rodopi Trakatroukides.

View Trakatroukides in a larger map
(Touchingly, the Greek Facebook page for those who have left Kızderbent links across to the Turkish Facebook page, for those who live there now.)

The Greek villagers of Kızderbent call it Κιζδερβέντ, Kizdhervent, and there is a copious variety of transliterations: Kizderved Kisderved Kizdervent Kisdervent Kizderbed Kisderbed Kizderbent Kisderbent.

And Къздервент. Because there’s something notable about the language of Maidenpass. It wasn’t Greek; or at least, it wasn’t just Greek. We have little left of their language—and I’d certainly never heard of it during my preoccupation with Balkan linguistics; but from the Motley Word, a site I reviewed a while back, we have some 500 words. (To see them all, click on Όλες in the interface.) A lot of the words are Slavonic, and a lot of the words are Turkish, and none are noticeably Greek. That could be a selection bias—the Motley Work lexicographer (who is still going!) may just be singling out the Slavonic and Turkish words. But from the evidence I’ve seen elsewhere (see the followup post), I doubt there’s a substantial amount of Greek vocabulary.

There was also discussion of the language of Maidenpass on a thread at Sarantakos’, which triggered this (and which I’ve already posted): the indication there was that there was no mutual intelligibility between Trakatroukika and Greek. From that thread, we also have a sentence of Trakatroukika, a song verse: tákar tákar ke priataláh duí sulísperi zebé khantúm

The indications are that the language of this village was somehow a mixture of Balkan Slavonic, and Turkish. At least one glancing mention, from the website of the municopality of New Triglia, implies Armenian was there as well, but I have not seen a direct confirmation. (I have seen something more indirect, on which see further down.) The Slavonic component is conventionally called Bulgarian, and certainly when Slavists visited Kızderbent in the 19th century, noone was making a distinction between Bulgarian and Macedonian; but there is at least one account saying the village was settled from Ohrid, and the language spoken in Ohrid is not now called Bulgarian.

The name the villagers went by, Trakatroukides, doesn’t give us much of a hint: as TAK anticipated in comments, it derives from traka truka, an onomatopoeia of firecrackers (as the Triantafyllidis dictionary gives it)—so “bang bang”. The onomatopoeia is also associated with grinding knives (carnival song: τράκα τρούκα τη μαχαίρα, a verse that has wandered into Rebetika referring to knives), and a children’s song about two steamships colliding (“The Greek boat has lemons. The Turkish boat has eggs. The boats go ‘bang gang’. The sea is avgolemono“).

The alternate name Rokatzides doesn’t help either: ρόκα means spindle or rocket, arugula (Italian rocca—hence also English rocket; or ruca). It’s tempting to associate “rocket” with “bang bang”, but I’d say that’s an anachronism—the association of spindles with rockets is late and Western.

The web site of New Triglia has this to say on the origins and language of the Trakatroukides of Kızderbent—the Bangbangers of Maidenpass:

There are three views on the founding of Kizdhervent. One claims that Greeks from Ohrid and Monastir [Bitola] migrated to Constantinople seeking a place to settle. The other opinion is that they did not migrate but were exiled because they robbed the nearby inhabitants. The third is that they were prisoners of Mehmed the Conquerer, who transported them to Constantinople and from there to Kis-Han, as he passed by Ohrid returning from Kroja. […]
Later on, because of the privileges the Sultan gave the area, Greek-speakers, Turkish-speakers, Armenian-speakers and Bulgarian-speakers settled the region, and that’s how Kizdhervent was created. The mixture of these peoples is also the reason why its peculiar language came about, spoken by no other village in Asia Minor. For that reason they were called Trakatroukides.
Their customs and religious habits are the same as other Greeks’, and their religion is Christian Orthodox.

The Bulgarian Wikipedia article has a somewhat different story, which is worth giving in detail:

Kăzdervent was the first village of Asia Minor Bulgarians attested by written evidence in the 19th century. [Accounts from 1807 on, almost all of which describe the village as speaking Bulgarian.]
According to stories by the Kăzderventsi the village existed in the 18th century, and is one of 12 villages donated by the Grand Vizier to a lady in the seraglio: she sold them to an Armenian banker, and he sold them back to the Turks one by one.
Konstantin Jireček later wrote of Kăzdervent, and inferred that they originated from Momim Prohod, since that is what Kăzdervent means in Turkish. According to Jireček (1899) the villagers “speak Turkish and half are already on the way to losing their nationality entirely.” Vasil Kănchov writes that the village had 440 houses, and believes that the Kăzderventsi were involuntary migrants from Ohrid.
At the end of the 19th century Kăzdervent was gradually Turcisised linguistically and Hellenised in national consciousness as a result of intermarriage and influence from other Karamanlides’ villages.

So we have in Kızderbent a village which spoke Bulgarian (or Macedonian Slavonic); which ended speaking something between Bulgarian and Turkish (somehow), and was notorious in its region for “talking funny”. (We have a report the Greeks and Turks in the area thought so; did the Bulgarians?) By 1914 however, the villagers did not consider themselves Bulgarians or Turks, but Greeks, and did not follow the Anatolian Bulgarians’ exodus from Turkey; they left Turkey with the other Greeks in 1922.

The history behind that is obviously complicated, and needs a separate post. Here, I’m going to pose some linguistic questions. I don’t have the answers (although I do hope someone who knows their South Slavic does go through the Motley Word list, and the language samples from the next post). For all I know, there have been Bulgarian studies of Trakatroukika; but I’ll speculate anyway.

The most important question is, what sort of a mixture was Trakatroukika? If the situation was something like the New Triglia article paints, with Armenians, Greeks, Turks and Bulgarians cohabiting—and no single language dominating, we might end up with a pidgin. Pidgins arise as contact languages all the time, between people in recurrent but not close contact. For such a pidgin to become the main language of a village though, it would have to make its way into the hearth. That would mean routine intermarriage (possible for the Greeks, Bulgarians, and Turkish-speakers if they were Christian too).

It would also only make sense if multilingualism was impractical. That happens on slave plantations perhaps, which is how pidgins become creoles. But it didn’t happen in Papua New Guinea, because the hearth can be multilingual; and although it may not now look like it, the peoples of the Balkans used to be multilingual as well. So the drastic breakdown of language which you get with pidgins—and the rebuilding you get with creoles—seem to me unlikely for Kızderbent. We’d know whether that had happened of course the minute a Slavist sees an actual sentence of Trakatroukika (like the one I pasted above). A pidgin or a creole would have very little left of the morphology of Bulgarian, and even less of its syntax.

True mixed languages, like Michif—the verbs are Cree, the nouns are French—are even more rare, more spectacular, and certainly more controversial. But they don’t result from language breakdown, but fluent multilingualism: enough villagers speak both languages that not only words, but grammar ends up seeping between the two. It’s worth noting that Wikipedia’s list is short, but includes Cappadocian Greek—because Cappadocian was not just borrowing words from Turkish, but vowel harmony and inflections: elements of the basic machinery of Turkish.

That said, mixed languages are rarely stable anyway, and at least some of them (I’d say including Cappadocian) are an attention-grabbing twist on language shift—language death, if you prefer. That’s certainly how Jireček read Kızderbent: their Bulgarian was mixed with Turkish, not because it was finding a new equilibrium, like Mbugu (Cushitic vocabulary, Bantu morphology) or Erromintxela (Romany vocabulary, Basque grammar). It was mixed because it was yielding to Turkish. Of course, if Trakatroukika really was like Michif or Erromintxela, Jireček may not have realised it, and I don’t know if we’d have enough data now to work it out. This was the time of duelling aphorisms on mixed languages (Müller 1875: “There’s no such thing as a mixed language”; Schuchardt 1882/83: “There’s no such thing as a completely unmixed language”), but linguistics was only starting to be sophisticated enough to realise a mixed language when it saw it.

A scenario like Cappadocian though makes sense, and I suspect (based on no real information) that that’s what happened in Kızderbent as well. The village language in SW Cappadocia has a Greek core, but is bilingual: several villages had both Muslim and Christian inhabitants (in one in fact the Muslims spoke Greek), and Cappadocia was also home to the Turkish-speaking Christian Karamanlides. The Greek-speaking villages themselves were shifting towards Turkish: the presence of Muslim Turks and/or Karamanlides made it that much easier. Because Turkish was gaining ground, and everyone was fluent in Turkish, there are more and more elements of Turkish grammar, and not just Turkish vocabulary, in the language the village spoke. In the Balkans, that same bilingualism meant that the grammars of the Balkan languages ended up looking fairly similar, even if their lexica did not. In Cappadocian, it was read—probably correctly—as a precursor to Greek yielding to Turkish correctly. Substitute Greek with Bulgarian, and what you get sounds plausible, at least.

To put the Cappadocian scenario in play in Kızderbent, it would help if we knew that the Turkish-speakers were on intimate terms with the Bulgarian-speakers. The Bulgarian Wikipedia (building on Kănchov? on that 2001 study on Anatolian Bulgarians?) thinks it was Karamanlides, who were Christian and so could intermarry. I think Bithynia is a bit too far West for Karamanlides, but obviously something Turcicised the Trakatroukides’ language.

Outside the main question, of how Trakatroukika actually worked, are the historical questions of what went into it. For starters of course, the question that Jireček and Kănchov worked on in their visits to the region: was Kızderbent’s Slavonic closer to Momim Prohod or Ohrid? (Or for all I know, a mixture of the two—although I presume these settlements were block moves and not mergers of multiple populations.) From the Bulgarian Wikipedia, I see we know a lot about Anatolian Bulgarian in general, and that it was mostly southern Bulgarian; it was possible to pinpoint some villages to particular towns—including Stengelköy, settled from “Kostur in Southwestern Bulgaria”. (The hyperlinked town on the page is Kastoria, now in Greece, and its Slavonic has not been called Bulgarian for a while; but there are a couple of Kosturs in Bulgaria as well, so I’ll assume lazy hyperlinking. 🙂

It’d also be worth checking whether there is actually any Armenian in the language mix, as New Triglia’s site hints. Bithynia is next door to Phrygia, and Phrygian is suspected of being related to Armenian—or a missing link between Armenian and Greek. Phrygia is a long way from the Armenian heartland, but Armenians did travel west during the Ottoman Empire; and the Wikipedia listing of Armenian dialects, from Acharian’s 1909 Classification des dialects arméniens, includes Nicomedia (İzmit) and İznik (Nicaea). So Armenian was spoken in Bithynia at the time (and there’s confirmatin next post too). The creation story of the village reported in the Bulgarian Wikipedia refers to Armenian involvement.

New Triglia also drops the hint of a Greek contribution, both in the four constituents of the village, and in its reference to “Greeks from Ohrid and Monastir”. (If your reaction is, “*what* Greeks from Ohrid and Monastir?”—well, hold that thought.) As mentioned, there is no evidence of mutual intelligibility with Greek, and the Motley Word lists don’t feature a single word that looks more Greek than Bulgarian and Turkish. The Motley Word is a Greek dialect wiktionary, so Hellenic words should have still been distinctive enough to include. (Of course there’s a selection bias, and I’ll come back to that.)

As a language of culture and trade (and the church), Greek had pervasive influence through the Balkans and into Asia Minor, including where Trakatroukika was spoken; and Greek was spoken by other communities in Bithynia:

• Danguitsis, C. 1943. Étude Descriptive du Dialecte de Démirdési (Brousse, Asie Mineure). Paris: G.-P. Maisonneuve.
• Δεληγιάννης, Κ.Π. 2002. Κουβουκλιώτικα: Το γλωσσικό ιδίωμα των Κουβουκλίων Προύσας. Αδελαΐδα.

From the Bulgarian Wikipedia again, we know both Turkish and Greek were encroaching on Anatolian Bulgarian in general, not just in Kızderbent: “Surrounded by many Greek and Turkish villages, the Bulgarians were forced to adapt and to learn the languages of the majority. According to the recollections of the older sons of Anatolians, almost all young and old settlers in Anatolia also used the Greek and Turkish languages, though many of them were illiterate.” But if Bithynian Greek had an effect on Trakatroukika, I suspect it would have been apparent to me by now.

And finally, “*what* Greeks from Ohrid and Monastir?” Well, that’s a messy question, and it’s best left for The Other Place, where I take all discussion of complications in what being Greek (or Bulgarian, or Macedonian) is about. You can probably already predict what I’ll say: the current Trakatroukides of Valtotopi or New Triglia identify as nothing but Greek—so they’re nothing but Greek. The indication is that they share a bloodline and a linguistic heritage with people who have never been Greek. And one must have a very restrictive notion of national identity, to think that a paradox.

Because of course in the 17th century, or whenever Maidenpass was settled by the Bangbangers, there was no Bulgarian or Macedonian or Greek identity as we now know it; there was just the millet-i Rûm of Orthodox Christians, which encompassed Greeks, Albanians, Bulgarians, Arabs and Romanians alike. People knew they spoke different languages (and very different dialects within those languages), but that was not how they defined themselves: that came later.

So yes, it may be wishful thinking to say that settlers from Ohrid in the 17th century were Greek—particularly as Greek that far north was an urban-only language, and Kızderbent is unlikely to have been settled by townsfolk. But if you’re going to deem them Bulgarian or Macedonian, and not just Slavonic-speaking, you’ll need a very big asterisk; and to their descendants, such an identification is irrelevant. (If you make it alive out of the argument over whether they were Bulgarian or Macedonian—which goes to show how immaterial the question was back then.)