Kozani: a stab at etymology

By: | Post date: 2009-09-22 | Comments: 18 Comments
Posted in categories: Linguistics, Modern Greek
Tags: , ,

Language Hat asks in comments to the previous post about the Wikipedia etymologies of Kozani:

According to prevailing opinion, the name comes from the village of Epirus Kósdiani, the origin of settlers of Kozani in 1392. The settlement was first named Kózdiani, which then, it was changed into Kóziani, and in the end into Kozáni.[2]

The name “Kozani” probably may also derive from the South Slavic kožani < koža ‘skin (goatskin)’.[3] The name of the city in South Slavic languages is Кожани (Kožani).

Am I wrong in thinking the Slavic etymology makes much more sense than the first one, with its strained reshapings? (We won’t even get into what “probably may also” is supposed to mean.)

As often happens in Wikipedia, it’s two opinions barely reconciled together on the same page. Now, Googling is no surrogate for actually knowing Slavonic (or Albanian), but let’s see how I can help. (Damn you, Language Hat, I was supposed to be doing my taxes!)

The “Hellenic” etymology is sourced from the website of the city of Kozani, which as you’d expect wouldn’t be eager to point to the parallel with that language. (Kozani prefecture turns out to be the furthest spread south of Slavonic in Modern times, with Greek and Slavonic villages mixed.)

[Non-Greeks will notice that I speak vaguely of Slavonic, when we all know *which* Slavonic language I’m talking about. But I’m not feeling like getting into needless argy-bargy with those of my readers who don’t want to call it “Macedonian”; and since I’m talking about 1400 and not 1950, I may just get away with it…]

The thing is, Kozdiani may not look like Кожани, but it doesn’t look particularly Hellenic either. Is there any way Kozdiani could originate from the same Slavonic form as Kožani? Although given it’s Epirus (Northern Epirus, as it turns out), and given the names of the other villages cited below, Albanian is more promising as an etymology.

This is what the Kozani city website has to say:

In 1392 colonists from Premeti, Bithikukio, and Kozdiani of Epirus {Përmet and Vithkuq, Albania}, fled chased away by Muslim Albanians to the region north of Selitsa {Selitsa, now Eratyra, Greece}, which is to this day called Old Kozdiani, and subsequently migrated eastwards, encountering the Christian settlement of Kalyvia. {The quite Hellenically named “Huts”; I’ve found one near Konitsa, on the Greek side of the border.}

The inhabitants of Kalyvia did not reject them, but they obliged them to build their houses further east. The new inhabitants called the region Tzamouria {i.e. Çamëria), preserving the name of their old region.

Nowadays the region is called Tzambra. {[dzamurˈja] > [dzamˈrja] by high unstressed vowel deletion > [dzamˈbra] > [ˈdzambra]}. They called the rocky hill over Tzambra Skrika or Skirka (Sk’rka), which means a rocky elevation.

Though there are different opinions of where the city name came from, the dominant opinion is that these colonists from Epirus called the new settlement Kósdiani, which then became Kóziani, and later scholars [i.e. in Puristic Greek] transformed it to Kozáni.

A few years later families from Servia {town in Greece} and Drepano move to Kosdiani, augmenting it.

From another site: “People speculate that the name Kozani is due to either the place of origin of its first inhabitants, Kosdiani or Kostiani in Epirus, or their main occupation of tanning: “coza” in Epirot means goatskin.” Tellingly, coza is given in Latin script, which you’d be unlikely to do if “Epirot” here meant “the Greek dialect of Epirus”.

The account derives from a 1924 history of Kozani, cited at this blog:

The name of the folklore group (Kóziani), stressed on the antepenult, shows a hardy genuineness, like the hulls of Sk’rka (Slavic: rocky protuberance; Albanian: hill). And given the opportunity: the definitive interpretation of the name of the city is stilll under research. I’m opening up the History of Kozani by P. Lioufis, Athens 1924.

  • The first inhabitants of the region one day saw … a she-goat running through the trees, and called the village Kózani or Kóziani; for kóza means a she-goat, and kózia a skin in “Bulgarian”.
  • Above the town of Selitsa (Eratyra) there is a region in its mountains called Kóziani, or rather Old Kóziani, where those pursued from Northern Epirus first settled, from the villages there of Kosdiani and Bithikukion. (The latter exists to this day.) Then they came to the region of Paliospita where the Kasmirtzidis Society have their establishment, as do many other settlers in the area, genuine or not.
  • The Turks called the town Kózana, from Koz (“walnut”, or Kákhta in pure Kozani dialect) and Ana “mother”, for the multitude of walnuts there.

And here’s a map of Përmet, Vithkuq, Kalyvia, Selitsa (now Eratyra, near Askion), Kozani, and Servia:

View Kozani in a larger map
At least one site says that the original Kozdiani was destroyed, which is consistent with Lioufis’ wording

So what do we make of all this?

  • The Turkish is a folk etymology, and nothing wrong with that.
  • Kozáni may be named for Kózdiani, but it’s quite possible that both are named for the Slavonic for “goat”; Lioufis certainly assumed so.
  • Kózdiani is in Northern Epirus, i.e. modern Southern Albania; Albanian and Slavonic are both possibilities for etymology, and Albanian and Slavonic would both have been trading words anyway (as appears to be the case with “Sk’rka”)
  • Tzamouria i.e. Çamëria is the name of Southern Albania, although we don’t know from this whether we actually have a record of the area around Kozani being called that at the time—which would confirm migration from Southern Albania—or whether this was Lioufis’ etymology of modern Tzambra. All of the Tzamouriá > Tzámbra etymology is plausible, except for the stress shift.
  • I’d like to know more about Lioufis’s sources, because the Greek Wikipedia article then adds that the first written reference to Kozani is in a firman of 1528.
  • I think 1392 is a little early for religious conflict in the Western Balkans—the reference is to τουρκαλβανοί, Muslim Albanians not τούρκοι. But I could be wrong. *shrug*
  • “The inhabitants to Kalyvia did not reject them, but” they kicked them out anyway.
  • I’m not at all convinced that the stress shift of Kóziani to Kozáni is a learnèd thing. Sure Kóziani violates ancient accentutation (because Κόζιανη looks like being stressed four syllables from the end—until you realise it’s [ko.zja.ni], but Ancient Greek did not have ι as [j]). But I’d have expected a learned form to go with Kozíanon, rather than preserve the feminine gender. And as Don Hat correctly noted, phonologically the shift from Kósdiani to Kozáni does seem a little forced.
  • So the derivation of Kozáni from Kósdiani has problems. But as it turns out, both Kozáni and Kósdiani seem to have a Slavonic origin anyway, so it’s a distinction that doesn’t matter.

So, that’s what googling tells me. If anyone actually knows any Albanian, Macedonian [there, I said it], or Greek dialect and can contribute, you’ll be doing the world a favour. Particularly if they can explain the accentuation of Kozáni.


  • Anonymous says:

    "Kozani is spelled out with a diacritical over the z, which indicates that the Austrian surveyors saw it as still predominantly slavophone"

    Considering that in the early 20th century, before the Balkan Wars, Kozani was a mostly Greek-speaking town (along with fewer Turkish speakers) surrounded by Turkish-speaking villages, perhaps a different explanation is in order. When even Vasil Kanchov portrays the town as Greek and Turkish speaking (though I'm not sure about his numbers), you know your theory is bad.

    Btw, in many areas of Macedonia, and I believe other areas of northern greek dialects, Greek zeta as pronounced as zh, though I'm not familiar with Kozani. I'll have to take your proclamations about the modern pronunciations with a pinch of salt. Cheers.

  • Language says:

    *runs, hides*

  • opoudjis says:

    … Well, I assumed Pierre's "Ko'zani" meant IPA [koˈzani] = Kozáni, not Kózani with apostrophe as acute à la TeX. So I quite misunderstood his evidence!

    An old Kózani pronunciation means the connection to Kózdiani is not problematic (although we'd still need see proof of that connection). Why the latter-day stress shift, though, if that's what's happened, is a real puzzle. Yes, Κόζανη in ancient accentuation is impossible, but I just cannot believe that kind of pedantry would have motivated this stress shift.

    In a certain language spoken immediately to the north of Kozani, we note that: "Other than recent loanwords, word stress in Macedonian is antepenultimate […]. By comparison, in standard Bulgarian, the stress can fall anywhere within a word." That would explain Kózhani alright, but not Greek Kozáni. If anything, it corroborates Tasos: "Kozáni" does not sound more "Greek" than "Kózani", but "Kózani" does sound more "Makedonski" than "Kozáni".

    This is getting more and more confusing. I blame Language Hat again! 🙂

  • opoudjis says:

    Especially for you Peter 🙂 , I am now highlighting Wikipedia links in the blog with a Ⓦ . It's a fairly old CSS trick by now, although it's not clearly to me whether Internet Explorer deals with it yet.

  • TAK says:

    Although I did not want to bother you all with another long comment, I have to make a few extra remarks.

    Pierre is right about W. M. Leake: on pp. 297-301 of vol. III of his Travels in Northern Greece , London, 1835, there are several occurrences of Kózani.

    That said, Leake is not always accurate on names/info. E.g. on p. 301 he writes: "George Sakellário, translator of a part of the Voyage D'Anacharsis and some other works, which he undertook for the benefit of his countrymen […]. His brother-in-law, Papa Kharísmio, who is now residing at Kózani, is an author also, and has written a Pantheon for the use of the schools of Greece."

    "Papa Kharísmio" is of course, Charisios Megdanis, the scholar to whom I have referred many times in my previous comments, who was Sakellarios's father (not brother) in law (Sakellarios after the death of his first wife got remarried to Megdanis's daughter Mitió, a well-known (to specialists again) scholar, who is in fact the first translator of Goldoni's plays in Greek. Leake's confusion may be explained because Sakellarios and Megdanis were roughly of the same age, born in 1765 and 1768 respectively. But "Kharísmio" appears to be inexplicably mistaken).

    However, as I said, in Leake there are many occurrences of "Kózani" and I take it for granted that this is the only name that he heard and he records it correctly. As I take for granted Pierre's statement that when he was in the region in 1960-1 all the Slavophones and the Hellenophones that he met called it Kózani.

    This, on the other hand, tells us little, or nothing, on the placename's etymology. Because as Nick said, we have to make sure that -an(i) is meaningful in Slavic to justify a Slavic etymology from Kóza (and from what I could find from readings and friends it is not – on this an expert's opinion would be useful and is still required. But if, according to all indications, it is not, then we have to accept that the etymology from Kóza is a folk etymology).

    Additionally, from what I know "Kózani" was never generalized among Hellenophones (it is not today and it was not in the early 1990s when I spent a couple of months there as a soldier), and in written Greek sources from the 18th c. onwards the name is exclusively recorded as "Kozáni": you may find examples of such sources in Κοβεντάρειος Βιβλιοθήκη , though the quality is not always great. I restrict myself to two easily readable examples: the first is Megdanis's manuscript that I have already mentioned (the link will take you to p. 6 where Κοζάνη is easily discernible). The second comes from another important local scholar of the period of the Greek Enlightenment, Michail Perdikaris (1766-1828) and it is a manuscript dated in 1805 (the link will take you to the title page where again Κοζάνη is easily discernible).

    Even if I accepted, for the sake of argument, that in all the period from 18th-20th c. the city was generally called "Kózani" and written "Kozáni", I would still have to explain why this was so, and how, in linguistic terms, "Kózani" became "Kozáni". Does "Kozáni" sound more "Greek" than "Kózani"? I seriously doubt it.

    And why would Megdanis and the other local scholars would call it Kózani when a foreigner (e.g. Leake) was present and Kozáni in their writings? Could it be that Kózani was perceived as the foreign name for the "Greek" Kozáni? Could it be that Kózani was closer to the internalized accentual patterns that foreigners had? Is it a case of 'phonological deafness'? Dunno, but I find it strange.

  • opoudjis says:

    Btw, Pierre, thank you for your contribution, and it is some evidence that Kozáni/Kozháni is independent of Kózdiani…

    If we do meet next month, btw, let's agree not to talk about Unicode. 🙂

  • opoudjis says:

    … By "scaffolding", btw, I'm thinking almost in Semantic Web terms (a URL for every concept on Earth, to anchor hyperlinks to). But again, as definitions rather than primarily as sources of argumentation. The argumentation happens in papers (which are not online), and in comment threads…

  • opoudjis says:

    @Peter: If I mention a Greek town, or variant of Arabic, or monastic author, I'll link to Wikipedia as a handy glossary. Is it academically reliable? Not always for detail or niche work (then again, I don't think the City of Kozani website is intrinsically more reliable either). Not never either. But it's useful as scaffolding. And it's also useful because it exposes to people the contentiousness by which we come to knowledge. (As this thread also illustrates.)

    @TAK: Mm. I still don't know: we'd really need to see what Lioufis knew about population movements. Kozdiani wouldn't also have been named for Cossana, and if Kozani was indeed settled from Kozdiani, it's hard for me to think they're not related. (Lots of "if"s there.)

    But -an- is indeed meaningful in Latin, and what I'd like to know now is, is -an- similarly meaningful in Slavonic or Albanian? If not, the Beneventum ~ Velvento parallel is compelling…

    I had the impression Sourdi was soldiers' slang; so it's local? Hm. Too far from Vlach populations for that to be a latter-day explanation?

  • Peter says:

    Hi, Nick.

    Do you consider Wikipedia to be a reliable academic source? I am asking because I've noticed that you cite it often.

    P.S. I can vouch for this: Πτολεμαΐδα has some of the most beautiful women in the country.

  • TAK says:

    However, I have to note:

    1. We do not know for certain that the 'habitation was interrupted in Byzantine times'. This is an assumption. True, the city is not recorded in Byzantine sources, but this does not necessarily mean that it was completely abandoned. It could have well been reduced in population and importance to yet another mountain village in Macedonia that deserved no particular mention in imperial sources. This is important, because all the etymologies you found are based on this assumption, which lacks solid proof!

    2. The Turkish etymology is a folk etymology, as you noted, but in my view, so is the Slavic. I mean, come on, who can really buy this: 'the first inhabitants of the region one day saw … a she-goat running through the trees, and called the village Kózani or Kóziani; for kóza means a she-goat, and kózia a skin in Bulgarian'.

    3. Both etymologies from kóza/kózia and from kózdiani/kóstiani present serious linguistic problems: kózdiani/kóstiani requires a change of zdi/sti to z and a shift of the stress from the antepenultimate to the penultimate, which is very unusual and highly unlikely. The latter is a problem for the etymology from kóza/kózia too, which additionally presents another, major in my view, problem: where did the -ni ending come from? If the story with the goat or any relation with the Slavic kóza/kózia is valid in this case, why wasn't the city simply called Kóza/Kózia? In my view, the most probable assumption is that the place name Kozáni was already there when the Slavs came and they invented a rural legend relating the city name with a word that is very common in their language and thus creating yet another folk etymology (the Ottomans did the same some centuries later). As you said, there is nothing wrong with folk etymologies (most of them are really amusing), but that's a completely different story.

    4. I discussed Megdanis's etymology with Tassos Karanastassis and he was quite surprised: he told me that Thavoris (Αντώνιος Θαβώρης) has proposed a similar etymology from Latin Beneventum (today the Italian city of
    Benevento in Campania) for the nearby village of
    . I couldn't find Thavoris's paper on the history and etymology of Velvendo, but, much to my surprise, I found
    this . So, on the official site of the village, Charisios Megdanis, Antonios Thavoris and other scholars who have favoured the Latin etymology are mentioned (which most probably means that the person who wrote the text for the site did her/his reading, cause I assume that the info provided come from Thavoris's scholarly work). I went back to Megdanis ms. and indeed he mentions Velvendo too – remember that his main point was that when the Romans conquered the region they (re)named many places and the proof for this is that not only Kozani but also many other places, villages, etc. in the region have names of Latin provenance.

    5. The way I see it now, the etymology I found in Megdanis and proposed here as only an alternative, seems to be the most likely etymology of the placename.

    6. To make it a bit more complicated: Kozani is also known to locals by another popular name that is Surdia and its inhabitants are also called Surdi! In its case, we have not one but two names of most probably Latin provenance that have not yet been adequately explained…

    Take care,


  • TAK says:

    Some additional things (in 2 comments cause they exceed the limit imposed by blogger):

    First of all, I think I managed to decipher the alternative name for Cosa that Megdanis provides in the ms.: it's Κόσσανον . This is indeed a Roman city in Calabria, the city of
    . [The pronunciation -z- or -s- is not a problem here; both could be valid depending on the dialect, e.g. Cossano Canavese is in piemontese Cusàn, etc.]

    I cannot exclude the possibility that Megdanis was guessing; it was a good guess though! As I cannot exclude the possibility that as a local scholar of the 18th c. he had in hand some info that we lack today.

    The important thing in this case is to check the historicity of his story – sth. that I cannot do.

  • Pierre A says:

    Diana (aka Nauplion) passes most of your blogs to me from across the hall, expecially when they concern the Nomos region of Dhytike Makedonia (I too can be cautious in this cactus field.)

    I know Ko'zani from the almost 6 months I spent walking in the region in 1960—1961 and, when I was there, the dwindling number of Slavophones (they were being persuaded that it would be better for them to move further north) called it Ko'zhani and the Hellenophones called it Ko'zani. That would seem to me to be one of the best arguments for the accent.

    W. M. Leake, bless him, gives a stress accent for all the contemporary placenames he records, and in his day it was Ko'zani. Felix de Beaujour might also be helpful.

    But the best source if you need to use toponyms for linguistic history, is the Austrian General Staff maps created in a resurvey between 1890 and about 1915. The surveyors had no loony chauvinist ideas about what they were doing (at least not in the southern Balkans). They recorded what they heard from what they judged to be the majority population of the location. These maps are the last honest historical record from the late Ottoman period and illustrate the delightful “Macédoine” of settlement that survived until the ethnic cleansing that folllowed the first Balkan War.

    They can be seen at


    You don't need to know Hungarian to use this beautifully organized site. Simply find the sheet you need on the general outline and click on it. There is a magnifying glass provided so that you can look at the part you want. Kózani is in the upper right quadrant of sheet 39-40 Joannina. Kozani is spelled out with a diacritical over the z, which indicates that the Austrian surveyors saw it as still predominantly slavophone. Unfortunately, they did not add stress accents.

    Pierre MacKay

  • opoudjis says:

    @TAK: you r0x0r. The scan of the Public Library of Kozani doesn't r0x0r, so I'll take your reading as read.

    Megdanis could be guessing; that's my hunch. Problem is, so could Lioufis, unless I see his sources. We know there was an ancient town on the site, but that habitation was interrupted in Byzantine times, so a survival of the old Roman name is—not impossible, but less likely.

    It's an alternative alright, and thank you for doing the donkey work (man, you r0x0r). Better research than *I* undertook! I'm still assuming Kozani is Slavonic though (/kozia/ > /koʒa/).

  • opoudjis says:

    MUST LEARN NOT TO SKIM!!! The English Wikipedia page points out that "In the south-west of the modern city, on Siopoto hill, there was a settlement named Kalyvia, between 1100 and 1300, traces of which are still preserved." So that solves that: ignore the modern Kalyvia I found much further west…

  • TAK says:

    Now, I cannot really tell you to what extent what I found is true, but you might find it interesting as an alternative at least.

    After I read your post, Nick, my first thought was that we should find a text as old as possible that would deal with the history of the region. Since I knew that Kozani has a very good library Κοβεντάρειος Δημοτική Βιβλιοθήκη Κοζάνης, I thought I should try my luck there.

    I searched the Online Catalogue and I found a history by a well-known (to specialists at least) local scholar of the period of the Enlightenment in manuscript form. The scholar is Charisios Megdanis (1768-1823) and the text may be found here

    Of course, the fact that the library has digitized part of its rare books and manuscripts was a nice surprise – the nasty part was that the quality of the images is so low that the manuscript could hardly be read…

    Anyhow, what Megdanis suggests in page 14 (according to the hand-written page numeration) of the ms. is that the name comes from Latin and that the city was named after a Cosa in Calabria in Roman times. Unfortunately, no matter how hard I tried I could not really read the whole story he narrates in pp. 14-17 due to the low quality of the scans. But the general idea is that when the Romans conquered the region they (re)named many places and the proof for this, Megdanis claims, is that not only Kozani but also many other places, villages, etc. in the region have names of Latin provenance.

    Now, I looked for a Cosa in Calabria, but I couldn't find one. I only found Cosenza . However, founding a colonial city named Cosa would not be unheard of for the Romans – check the case of Cosa in Tuscany . What is even more interesting is that a man from Cosa in Latin would be called Cosanus (and of course the name is very well known and attested even today in the form Cosano in Italy). If there is a grain of truth in the story Megdanis tells, then perhaps a connection with a Roman Cosa (or a Roman Cosanus) should not be excluded. Especially if we take into account the fact that the genitive of Cosanus is Cosani [= Kozani!].

    Perhaps the Roman name of the city would have been sth. Cosani [in which case you already have the accent on the penultimate and you do not need to bring in any Kózdiani or Kóstiani or goats and goatskins…].

    Archaelogical excavation has brought to light a cemetery of Roman times in modern day Kozani and the findings in the graves suggest that there was a rich Roman city-trade centre there (unfortunately, we do not know how this city was called; check this in Greek).

    Of course, I cannot check the historical truth of what Megdanis claims. I only offer it here as a possible alternative.


  • Nauplion says:

    There used to be a Calive outside Nauplion where — surprise — Albanians lived. There is at least one Kalyvia in the Peloponnesos now, according to my driving map.

  • Language says:

    Sorry to take you away from your taxes, but I deeply appreciate the research!

  • opoudjis says:

    The Kalyvia I found is far to the west of Selitsa, so it may not be the right one for the City of Kozani website account. Instead of Kalyvia "huts", Lioufis referred to a region of Paliospita "Old Houses", which is presumably somewhere completely different.

Leave a Reply

  • Subscribe to Blog via Email

    Join 291 other subscribers

  • October 2017
    M T W T F S S
    « Sep    
%d bloggers like this: