 I am, I'll be talking about North East Caucasian or East Caucasian that's basically the same thing. A language family with a couple of dozen languages in them. I am not a Caucasology as by training I am an Indo-Europeanist specialising in Celtic but I thought a few years back it would be interesting to look at a completely different language family in which very little comparative work had been done. Also a different language family in a very different you could say social linguistic situation because we're dealing in this case with a language family on a very small area on the map. A highly fragmented area on the map where the ideas that these languages have been around for a very long time and they've been beside one another for a very long time. Maybe in contrast to Indo-European which spread across typical spread zones this might be in the terminology of Joanna Nicholl's residual zone where languages differentiate but always remain in contact so there's also diffusion with possibly pernicious consequences for reconstruction. That's for example claimed by Dixon for Australian languages. Well let me very briefly introduce the language family to you if only in terms of mentioning a lot of names of languages. The North East Caucasian language family also called Nach Dagestanian consists of basically two branches Nach and Dagestanian. Nach is Chechen, Ingush and Butts. Three languages pretty closely related to one another. All of them north of the main Caucasus mountain range apart from Butts which moved over into Georgia so to the south. The Dagestanian language is what Dagestan is as you probably know a republic in Russia about the size of Bavaria but with more than 40 languages being spoken. Which falls apart into a number of subgroups and the subgroup I'll be pestering you with is the so-called Avar-Andi-Dido subgroup which consists of Avar which is a fairly big language with about three quarters of a million speakers which also functions as a lingua franca or contact language because everybody in Dagestan especially in the mountainous Dagestan area will be able to speak Avar beside other Dagestanian languages. So Avar is the only really big language with a quarter of a million speakers and the Andian languages, Andi, Achwach, Chamalau, Tindi, Bachwalau, Boclig, Karatau and Golubiri are all very small languages and so are the Tidow languages Consisting of a Western branch, Cez, Hinuch, Chwabshi, Ingwchwar and Eastern branch, Besta and Hunzib. Talking about size of speaker numbers, the biggest language among the Andian-Dido languages is Cez which has 10,000 speakers approximately. And the smallest is Hinuch, which is the language of just one village, namely the village of Hinuch, which has 200 speakers. And the others are somewhere in between. So that's the group, sort of the central Dagestanian group in the high mountain area bordering on Chechnya, which is the western most neighbour and Georgia in the south. That's the area where you'll find the Andi-Dido languages. And if you go further to the south in Dagestan, you'll enter other languages which also belong to this family like Lach with about 100,000 speakers, Darhwa, with almost half a million. And a large family called the Leschi family with big languages like Leschi, about half a million speakers and very small languages in that group as well. Like Kinalu, which belongs there, or Archie, which are basically the languages of one village each. That's the family. Now when I try to, some previous work has been done on the family, but it's not really what you would call following the comparative method, because most of the work that's been done on the family has been descriptive. And typological. And there have been some attempts to sort of reconstruct in one go, for example, from the modern languages to Proto-Dagestanian or Proto-Nach or Proto-Nach-Dagestanian, without much attention to intermediate stages, which is slightly problematic because the best guesses as to the date of the Proto language would be, between four and six thousand years ago, which is a hell of a long time. So when I tried to work on this language, it was quite interesting to see whether the comparative method would work or not. Well it works wonderfully. Aus nam sloz icheit all over the place. Those languages are also extremely happy to go through an enormous amount of sound changes. At the same time, and some of them shallow and some of them probably rather deep because they connect more than one branch. And the complexity resides in the fact that the sound changes are the, as you would know from in the European probably, the output of one rule is the input of a following rule. So relative chronology of your sound changes is a very important thing to establish. It's also very complex to find out if you know nothing. So a very lively historical phonology, comparative method, works a treat. Also these languages in the central area, so the Avar on the Dedo, they are morphologically very regular, which is bad for reconstructive purposes. Because that will mean if they have undergone lots of sound changes and are morphologically regular, you know there's been an enormous amount of analogical levelling. That's a certainty. What I wanted to pester you with today is not going into all the gory details of the language family, which is completely unknown to most of you, but focus on a few points, basically one point that struck me as an Indo-Europeanist and which I didn't really expect to find coming from an Indo-European background. And the first point I'd like to make starts from an example based on the Achwach language. Achwach is a language with 5,000 speakers approximately. It belongs to the Andean group. And Achwach is, one should distinguish between Northern Achwach and Southern Achwach. Northern Achwach is spoken in four villages in a pretty remote part of the mountains. Not that there are no near villages, but these villages are very difficult to get at because of the nature of the terrain. So basically you reach Northern Achwach through a broad valley and neighbouring languages would be high up on the mountains on either side of the valley and would be out of reach. So you go through the broad valley and then you enter into a river valley because most of these villages would be along rivers where you have four villages. And those four villages, they have no distinguishing features amongst one another so that they all of them speak the same dialect, which is extremely unusual in the Caucasus. And also the speakers of those villages have very little contact with people outside of their villages, which also is very unusual in the North Eastern Caucasus. Because especially men traditionally would travel quite a lot, first of all to get their women because these societies are generally exogamous. And secondly because they're usually cattle breeders and especially men would travel with their herds of sheep or cattle would travel between winter pasture and summer pasture and sometimes across mountain ranges. Sometimes with the effect that some of them wouldn't return. But so Achwach among North East Caucasus is in some respects exceptionally. So there's North Achwach, you reach that through a valley, four villages, same dialect and then you run up against a high mountain wall. If you were to cross that, which is pretty difficult to do, especially if you're a vertical, if you cross that you come into the area of Southern Achwach. Three villages with three dialects and they're rather different from one another. The Ratlub, which is closest dialect to Northern Achwach, but is geographically remotest. And Tseigop, and they live there, those speakers among speakers of Avar. And the linguist who did an enormous amount of work among the Achwach, both North and South, was a linguist, a Russian and Soviet linguist called Magomed Bekelra. And she travelled widely and did enormous amount of field work between 1948 and 1959, continued to do so later. And she has written the reference grammar, or to steal the reference grammar of that particular language. And she states that Northern Achwach people look very different from Southern Achwach people. Because Southern Achwach people, they look like their Avar neighbours. So dark hair, dark eyes, culturally very similar to Avar neighbours. And of course they would all be bilingual, but that's trivial. So all speakers of Southern Achwach would be bilingual because they would also speak Avar. Avar yn Achwach. Now Northern Achwach, they would also be able to speak Avar, but they look very different, she says. She says, there's a preponderance of people with red hair. So if you're looking for a reason why a person specialising in Celtic ends up in Northern Achwach, there's your answer to do with hair. So they're red hair, fair skin, light blue eyes. And as she says, Mother Rebecca, their women have a sharp tongue, which is very uncharacteristic for Caucasian speakers, she says. So this is the isolated position of Northern Achwach in Daghestad. And that expresses itself in some archasins shown in the Achwach language. I've given you the continental system of Northern Achwach, which is fairly extensive and also typical of this type of language in the world. I would like to point you especially to the difference between long and short consonants, or also a better frame, probably not long and short, but simple and intensive, because the long consonants usually have a larger area of occlusion to them. So they're not only longer, so you could say intense, which is pretty stable among those languages. A large amount of consonants in the lateral spectrum. So we have six voiceless laterals, short te, long chle, glotolised short te and long glotolised chle, and the two fricatives, chle and chle, and a similar elaborate system in the uvuller range, short aspirated chle, non aspirated long chle, as the long glotolised. Not so long ago it would be capable, you would be able to listen to the news in Avar, and Avar has those consonants as well on Radio Free Europe. So they would repeat news broadcasts in Avar, and it's particularly those long intensive glotolised consonants that jump out of the radio. So you wouldn't need to know, understand a lot of Avar to be able to recognise it in one go, because of these consonants, the long intensive ones. The consonants between brackets are suspect of being only present in borrowed words. But if you would take away the brackets, you sort of end up in the proto-Eastern Caucasian system of consonants. And it's to be the case. Now, Nordenachwach innovates a bit, so it's very archaic, but it also innovates. It innovates in having a reshuffle of sibilance and shibilance, so sibilance being the zy type sounds, and shibilance being the shy type sounds. So the short sibilance all become sibilance, and the long sibilance all become sibilance. That's what they do. Why do they do that? So it only occurs in Avar, northern and southern, sorry, in Achwach, in northern and southern, and it occurs in a particular dialect of Avar that is being spoken among the southern Achwach people. So it's likely to be a contact phenomenon, although it's very difficult to decide who started this. But this is not the thing I'm after. I'm after what's at the top of page two. The particularly unstable sounds in this very large system of consonants are the short affricates. So cle, cre, tse and she. But they behave extremely asymmetrically. So the cle and cre become affricatives in all Andean languages, in fact also in Avar and also in the Daido languages. So basically in all languages apart from northern Achwach, tse becomes a affricative of the type tse. In all Andean languages, in all Daido languages and mostly in Avar but not completely in Avar, whereas short tse is retained everywhere apart from the Daido languages. So completely asymmetric treatment of the affricates, the short affricates. But let's have a look at the two short affricates that in this huge language family in terms of diversity are only retained in northern Achwach. And that's the short cle and the short ch. So I've given you basically the data there. Ch becomes ch in southern Achwach but it remains ch in northern. Ch becomes ch in southern Achwach but it remains in northern. And let's not talk about tse here. Now, if it weren't for northern Achwach, so those four villages with a couple of hundred speakers in them, because all the other languages in the family have ch instead of ch. They have ch instead of ch. If it weren't for northern Achwach, we would no doubt reconstruct the sound change from the affricate to the affricative to proto Avar on the Dido. It's only because we happen to have northern Achwach that we can't anymore. And this is, I feel one of the typical things here. If you look at in European, the impression you get always is at least the impression I have is most of the stuff is gone. Most of the stuff in the dialect continuum is gone. So it's the equivalent of having say the romance language family as it is today. And the only thing we would be left with after 2000 years is the dialect of Paris, the dialect of Bucharest and the dialect of Madrid. So no messiness basically, no continuum. I think it's quite different in the Caucasus. It seems that all these intermediates or at least very many intermediate stages are still preserved. Now I want to take you through the developments of ch and ch in the Dido languages, but I do so extremely briefly. I've given you the material to read at your leisure if you feel like it or not feel like it, you can skip it. This is basically my etymological dictionary of those languages in the state of being and becoming. In the Dido languages, the ch has a fairly complex treatment in some languages, so not in all of them. So tle becomes sle, which is basically what happens all over the place in four languages. Ces, Hinoch, Gwarsch en Beista. I'm now page two, number five. But in the Inchokwa language, there's a pretty complex set of correspondences. Beginning of the word, the reflex is H, after a vowel it's L and after a consonant it's sle. The other complex language is Hunzip, which has sle at the beginning and after a consonant, but le after a vowel. That's based on the material I gave you on page two, on page three, on four, on five. Now I'm on page five at the bottom. At the bottom I again repeat the situation there. So if you look at what happened to Proto Avar on the Dido Club, short Africa, this is what happens in those languages. Now normally you would be inclined if you saw a change happening, especially across many languages, to use that in order to draw up a family tree. So to look for common innovations and then plot them and say more or less pinpoint when this development happened, by looking at which languages to part in it and use it in order to draw up a family tree. That is pretty difficult in this particular case, because there are four languages in which nothing very spectacular happens. You just get sle, the fricative. But there are two of them which do different things, especially Inho Kvar and Hunzip. But the isoglosses you can then draw through the family, they don't align with any deep distinction between the languages. So the deep isogloss, the isogloss separating Inho Kvar from the rest also separates the village, basically three villages of Inho Kvar, from the neighbouring village of Gwars, which is almost the same language. And it's the same with Besta versus Hunzip, they're very closely related to one another. The one has the trivial treatment to sle and the other has the non-trivial treatment. So you get isoglosses that do not conform to the genetic grouping you might expect. That I find slightly unusual. Well, anyway, what we can basically say on the basis of this reconstruction of the Daito languages is because Inho Kvar and Hunzip have a very specific and particular set of correspondences that the short African must still have existed in Proto-Daito. So it was lost earlier, you wouldn't expect this. So there's a unique correspondent set for Proto-Avar on the Daito kle in Inho Kvar and in Besta. Sorry, and in Hunzip. It follows from that that a Proto-language of the Daito language cannot yet have lost kle. It was lost definitely, but it was lost at a later stage and separately in each individual language. And that's again is something I would normally not be, I would not expect to find very often in European. Of course it's quite natural that this might happen. Language is doing the same thing independently of one another. But here you can actually pinpoint it that it does happen in this particular case. Sort of the same story that goes a bit quicker for kle, the usual. And again there's a bit of material and turn a few pages, you end up on page 11. There a set of correspondences is slightly more difficult, so I've given you a diagram there. But again it depends on the position in the word and in which particular language you're looking. What this short African will do. The word initial position, there's a correspondence set which is unique to kle. So no other sound has that correspondence set. I skipped the second line of that diagram. After a vowel, again there's a unique set of correspondences among the languages, which you wouldn't find for any other sound. And after a consonant there's a non-unique set of correspondences because after a consonant, short ge behaves like long ge and like the figurative ge. And I've given you kle in the same type of diagram. Below that I'm on page 11, number 7. So it turns out there's quite a parallel treatment between kle and kle in a sense that apparently it looks like this particular sound was preserved longest at the beginning of the word and after a vowel, whereas it was lost absolutely earlier after a consonant. Okay, now I'm almost there. The first conclusion from all this would be Proto-Gaido represents the same stage of the development of the short non-ejective African, so currently, as the modern Northern Achwach languages do. And what I find remarkable then is the second part of the conclusion having looked at the Proto-Gaido languages, it isn't the case that Proto-Gaido then innovated in some sort of way know all the individual languages at a certain point independently of one another, got rid of the short Africans independently of one another, it seems. Does that make sense or perhaps it makes sense because as I said all speakers of Daido would also be speakers of Arvar and in Arvar the distinction was lost. So if sort of Arvar phono-tactics takes over, you could imagine that there would always be a trigger nearby to get rid of the short Africans. It sort of makes sense. The first consequence on bottom of page 11 I've already gone through. I'd like to turn the page and that's sort of the last thing I wanted to do. That I now find it very difficult and again I have the impression that it's a symptomatic for this dealing with this particular language family in this particular social linguistic setting that it is very difficult to distinguish chronologically deep isoglosses from recent ones. So what you could imagine happening is to think of the Northern Achwach scenario. So why is Northern Achwach so conservative as opposed to the rest of the family? Well it could be that Northern Achwach was isolated at a certain point as it still is and the other languages were sort of mooring to contact with one another or with Arvar and it was a common innovation that got rid of and in the entire family apart from Northern Achwach because it was isolated. But then isolation stopped and Achwach took part in a whole host of developments that turned the Andean languages into Achwach. That's one possibility. It could be a deep isogloss, separation, not being the first step in the languages moving apart from one another but separation being just temporary, just being temporary and then the next phase, no separation anymore. So that's a possibility. The other possibility is that we're actually looking at a very shallow isogloss. So all Andean languages could have preserved Tle and Khw up to the 19th century before Baron Oesla came along to make the first grammars of them. And only relatively recently all of them got rid of those sounds perhaps under the influence of Arvar whom I have started it apart from Northern Achwach because Northern Achwach is and still is pretty isolated from all the other languages. This apparently is what happened in Dido where you can actually demonstrate the change to be recent and I'm disturbed by the fact that in Indo-European I normally have no problem in distinguishing an ancient isogloss from a modern one. I do have this problem in this particular language family. Since I don't want to pitch Indo-European against Northeast Caucasian I want to draw attention to a few Indo-European parallels to both scenarios. There aren't many of them as far as I know. So the first scenario, so deep isogloss and then unification again across the language area. You get something like that in the vulgar Latin so late antique Latin vowel split. The Latin vowel system collapses. You lose the difference between short and long vowels and they collapse in some sort of way, some sort of pattern. There's the Eastern Romance pattern and the Western Romance pattern. The Romanian would be Eastern Romance and French would be Western Romance for example. Fine. But that isogloss runs across Italy. So there are Italian dialects who have the Eastern type and there are Italian dialects that have the Western type. That looks to me like a possible parallel for the isoglosses old. There was a separation at some point but that separation was not the starting point for further differentiation. Language is moving apart. But later changes then simply took over the entire area and didn't really matter that there was a different vowel system and suddenly Italy as opposed to Northern Italy both of them became Italian. And something similar could be said for what I've called Northern Russian here. I mean the text of the Russian, sorry, the Slavic language of the Norfolkwrot birch bark. Texts which failed to go through the second parallelisation. Apart from that look like the pretty normal standard medieval Slavic. Possibility two, so the shallow isogloss. So for some sort of a reason all these languages are held on to the sounds up to a very recent station that independently of one another or seemingly independently of one another then decided to throw them away. There is a parallel for that in European as well in terms of the loss of the laryngeal sounds and almost no language preserves them. It seems as a conspiracy that almost right before written at a station starts they've gotten rid of their laryngeals. And something similar with the Slavic residents. Very few languages preserve them but they all get rid of them in different ways depending on the subgroup. So we know it from inner European that type has changed. We think it's exceptional. My impression is that in North Caucasus this is not exceptional but quite common. Thank you.