 I must first begin by thanking Nathan Hale for providing me with this excellent chance to present a little bit my, it's not really even work, my working progress before such, before an audience with such an irradiation, which I can only dream of. And so today I, as you know, I have been working on the problem of grades in Tangud for a moment. And so today I'll just pick up where I left off back in Hamburg where, because I didn't treat the thing about the grade two. And so today I'll talk about it and talking a little bit, I'll be also talking a little bit about Chinese also. So I call the, I call the late middle Chinese, the Chinese dialect that is spoken by Tangus and the Chinese dialect that the Tangus learn, or learn Chinese well. So let's start with some basic information. So the most interesting to me category, phonological category in Tangud is the start of grades. So usually, by usually I mean by Huang Cheng, usually we reconstruct three grades for a Tangud on the basis of a conjunction of rhyme, order, and dictionaries. So in dictionaries like the one have all these C characters. And we have other lexicographic materials that present the exactly the same order of rhyme, of rhyme categories. So it's something that is very much agreed upon by the Tangud philologists themselves. And the other thing is the Tangud Chinese transcription evidence. There is a slight asymmetry between the Chinese to Tangud side and the Tangud to Chinese side, but for the grade problem it's okay. So as an example, we take the rhymes R34, so R34 to R37. And we see that so the first rhyme in this group transcribes almost only Chinese syllables in the Chinese grade one. And the second rhyme in this group, R35 transcribes only Chinese characters of the grade two. And the remaining two rhymes, R36 and R37, more or less indiscriminately transcribes Chinese grades three and four. In fact, these two are in near complementary distribution. And if we look really hard at the near part, we see that they are usually artificial. And we can, for most of these groups like this, we can make a case for real complementary distribution. So I agree with Gong Huang Cheng in reconstructing only three grades. So the traditional reconstruction of the value of the grade is made by Sofranov basically as a mirror image of the Chinese reconstruction at his day. So he considered the grade thing basically a thing of mediums. So the first grade has no mediums. The second grade has some kind of weak yard. And the third grade has some kind of strong yard. This is carried over a verb by Gong Huang Cheng. So you have the Niesge-Geslawi Yoda represented by a volcanic E in Gong Huang Cheng's reconstruction and the stronger yard represented by real yard. This creates the problem that it's very difficult to pronounce EA and EA. What's the difference between EA and EA? Well, I can pronounce a lot of things, but this one is a little bit difficult. So Gong Huang Cheng made a decision to, he basically invented an aliphany where in grade three the vowel is slightly raised. So you have EA and EA. What is pronounceable? A little bit pronounceable. But sadly it is not supported by the transcription evidence. So my favorite example is just the basic mantra in Mahayana Buddhism. Om Manipal Nehum. And if we look at the usual reconstruction we have Om Manipal Nehum. Clearly it doesn't work and it needs to be updated. And I have the whose part to consider the updating basically done. So at the moment let's start with a little bit context. So Tanggut, it is known already at the time of Wolfen Deng and Wang Xinru, that Tanggut is a Changi language closely related to languages like Chang, to Minyak, to Garongic languages. But at this time those languages are not very well described. So basically in order to be able to do the nice kind of comparison you need to really do great dictionaries of those languages and you like thousands of words before you can really get to something. So now with Guillaume Jacques, who has worked for more than a decade on Japook, he has a very great dictionary of Japook and he has looked for all the possible coordinates and he found some 300 of them. So now we have something very solid to work on which is the cognates between Tanggut and Japook Garong. So based on comparison with Japook Garong, which I will not explain today, I happily and gladly redo the presentation that I have done in Hamburg and elsewhere if you want it. And so basically what I suggest is that there is a phonological distinction of uberlarization, so between uberlarized vowels which are pronounced like bah, bah, and non-uberlarized plain vowels which are pronounced bah, bah. So the distinction between grade 1 and grade 3 which were reconstructed before as bah versus biah is now reconstructed as bah versus bah. This distinction of uberlarization triggers an aliphony for the velar, so basically the yain initials between qa and qa. It is very clear that grade 2 is uberlarized as grade 1 but we don't know what the thing distinguishes grade 2 from grade 1. So for now I will write an R for grade 2 and we will have a look in detail. So where does the uberlarization in Garong come from? Well, the first origin is what I call the primary uberlarization. It must be reconstructed back to proto-Changgi and it has some correspondences within my other silent Tibetan languages but it's too early to say anything serious about it. However, if we just look at Changgi languages the thing is quite clear. So for example, for past long it's boh uberlarized in Tangut and in zhibu garong the language I'm working on is dazwu dazwu with the velarized or uberlarized vowel which I call emphatic vowel and we have the same thing in Minyang the language which shares the name of Tangut is ba ba. So for willow it's uberlarized in Tangut and it's emphatic in zhibu etc. So this is the primary uberlarization which is inherited from proto-Changgi we also have secondary uberlarization so basically the root wasn't uberlarized at the stage of proto-Changgi but now there are other things which went to the syllable and which makes it uberlarized. So for example, if we have a uberlar if we have a uberlar called emphatic vowel so for example to we which is ta in zhibu there is the ge and this ge goes to the syllable and so we have not la but la for snake it is the same ta but in the first position so basically we should have pu or something for snake but it's ha pu and this ra this uberlarized element goes on to the syllable and makes it pu and not only uberlarized but veeler elements basically veeler codus and em which I hypothesized as having turned into some kind of another word than to this ge and these also make secondary uberlarizations so what do we find here we find the ubiquitous pattern the basic conspiracy in the historic phonology of Tangu which Mark Miyake describes as compression I have changed his definition a little bit and for me what is compression compression is that materials phonological or phonetic materials around the syllables so for example the codus or pre-initials or even some kind of minor syllables they all went to the main part of the syllable and become some kind of pan-syllabic feature which works for the syllable as a whole that is for example how we usually explain the origin of retroflexes in Tangu so you have pre-initial R or codon R which just extends to the syllable as a whole and itself it has disappeared so this is the compression and so the uberlarization in Tangu fits very well into this great picture of compression so we have left the second the grade II unexplained it has something to do with R so I will write it as R in capital letters but what is it? first we will look at this transcription evidence for this great R so for the Chinese transcription there isn't much to be said first it transcribes grade II second it transcribes grade III with Zhuangzu shengmu so with retroflex fricte for African as an initial the Tibetan transcription is more interesting I think it's for the Hanbi and Hanbi and am I right? where we have the vowel R in the second in grade II transcribed as R so for example this character which is Ba is transcribed as Ba and you never find the first find the grade I syllable in comparable circumstances transcribed with an R that's all we have got for the transcription materials it's quite hard to interpret it's more interesting to consider the etymological origins of the grade II capital R so the first origin of the grade II capital R is eulerization plus medial R so for example if you have Ka without the R you have Ka in grade I but in fact because there is an R it's krak so the R is reflected as the grade II also for wele it's zhembr and it is duly reflected as Ba with this capital R for white here the eulerization is not primary but comes from the code M however because there is the R so code M brings eulerization and eulerization plus medial R brings grade II the same medial R a non-neuvelarized context just disappears so for example this is a word which is shared by almost every Changi language so it's quite useful gr which means sinew and it is there is an R but it has disappeared because it's not eulerized so the same thing for clear water which is angri and just gi there's no difference between gi and gr so that was the first origin for grade II in Tangut and now things get more interesting here we have the R which has an euler initial with the vowels e, a and u so basically if any of you speak a language with euler so you know that it's very difficult to pronounce pi it's almost impossible to do q with a normal e so there is always some kind of intervening transitional element between euler initial and high vowel or front vowel and however in Tangut this thing is categorized the same thing as the other origins of grade II that is why I think that Miyake's hypothesis that 104 and 204 are grade II rhymes and not grade I rhymes is correct because basically you have other you have other initials one and two and one but you have almost only in fact euler initials in 104 and 204 so the R thing here is some kind of transitional element which has a low vowel in one of the garantic languages which is a variety of crossgap spoken in Wozi we have the same thing so for difficult which is you have however a normal word with a in Japhook is just reflected as e in crossgap and grade III e in Tangut and why we have this in Rai is because it's difficult to say if you say in Rai and you don't pay very much attention it becomes in Rai the last the last origin for for the second for the grade II in Tangut is retroflex initials I think I don't have the time to explain it clearly here so I've just declared by Fiat that the the shivalence so the Chantu the Chantu initials are pronounced are pronounced retroflex before a euler before a euler rise Rai so what does it say it means that because we have because shivalence are only compatible with grade II but not grade I it means that the R the capital R is some kind of transitional element between retroflex sound and a vowel and any vowel so for those of you who would love to look at a little bit at the argument I leave it here and let's go to the summing up so in order to recapitulate the evidence about the capital R so what does the capital R feel like first there is some kind of throttiness or backness because it is only a euler rise context and it's a transitional vowel between a euler and a high or a front vowel so it must be something like R or O or A or something like that that's our first intuition the second intuition is that the sound has an R-like character so otherwise it's impossible to explain the R in the Tibetan transcription and it would be difficult to explain also the origins with in R and with retroflex however we also know that it's not a real R because first we have the middle R in Tibetan and it's never transcribed as a middle R in Tibetan also it's in the how do you call it in the in the mantras if there is a word like and it will never be and you never find a grade 2 character used to transcribe in fact you have very few grade 2 characters in mantras basically it's used to transcribe the long vowel and nothing else and thirdly if it's really R it's impossible to explain why it becomes grade it's not okay grade becomes grade it's strange so what do I feel about it well I think that it's the I call it the eye hypothesis because I think it is an eye so phonologically I will call it a pharyngeal and phonetically it's the vocalis vocalis correspondent of an eye which is the A and very pharyngealized so for example white is pronounced difficult is pronounced grade and to cross the river is pronounced down and why do I think so well first it satisfies the the throatiness because you can't get really you can't get more throaty than the eye it's almost the most throaty consonant available and it also counts for the R-like character this subject is seriously little treated in the literature but if you ask anybody who works on Caucasian languages or on Arabic and you ask them what do they think in American English and in Mandarin Chinese you have this kind of rotic vowels and they will tell you that they are not rotic the principle acoustic and articulatory in those vowels is the pharyngeal construction and we find everywhere this kind of this kind of affinity between roticness and pharyngeal construction for example in the Chinese transcript of Arabic if you have an eye then it's rendered with one of the rotic vowels in Chinese that's my theory about that's my current theory about the grade 2 in Tangut and let's just project it back to Hexi late middle Chinese well, it means that grade 2 has some kind of RTR quality and it has a pharyngeal so my hypothesis is that the iron hypothesis is not only true for Tangut but it also works at least for Hexi late middle Chinese now the question is this is this grade 2 iron in Hexi Chinese influenced or co-evolved or anything like that by Tangut or it's more widespread in Chinese and in fact the Tangut is influenced by the Chinese well on the nature of grade 2 in Chinese basically we have two theories the A theory and the E theory the A theory is almost correct because if you look at modern dialects and if you look at all kinds of information it's clearly something like A however how do you get from Ra to A it's very difficult on the other hand if we use this iron hypothesis and so if Chinese grade 2 is characterized by strong pharyngealization then it's very natural to have the vowel A which is in fact the pharyngealized counterpart of A associated with grade 2 one is that we have this association between retroflex initials and A in Tangut and it's also very projectable to Chinese because not only in Hexi Middle Chinese but also in the early Northern Mandarin contemporary to Hexi Middle to Hexi to Hexi Middle Chinese we have the same thing we have the whenever you have a retroflex African or African in grade 3 syllables it is classified together with grade 2 so not only original grade 2 is grade 2 but in late Middle Chinese or in early Mandarin or in fact in most kind of Chinese retroflex plus everything gives grade 2 and we still have some modern reflection of this in dialects from standard Mandarin to Sano-Vietnamese every time if you the original retroflexes cause a lowering of the vowel just put Cantonese here but it works for any other language for most other languages as well so I have argued for the I true nature of grade 2 in Tangut and I project it into Chinese but it's not in fact it turns out that it's not dialect of Chinese but very common Chinese and the only thing it gives us is that the Tangut treatment you will arise into this kind of vowel from Chinese why do I say it because in modern Changi language you never find this treatment this particular treatment it's either preserved or lost without any trace or become some kind of real celebrity or it triggers the particular Tibetan sound change which brings gr to zh for example for the worst sign you have in northern in northern which is the select big vertices treatment and we have the Tibetan treatment zh in southern and this if you look at the dialect map the treatment are really scattered so it means that in most of the Wuhan map of Changi the middle R is preserved to a very late date so the most plausible scenario that Chinese all kinds of Chinese always had some kind of wine for grade 2 and Tangud has developed under the influence under the influence from Chinese so in our final slide can we speak about Heizhish Park I got interested in the first because I thought it would bring to some very clear area so Hexi Chinese is influenced by Tangud and Tangud is influenced by Hexi Chinese however I think Tangud is very strongly reshaped, restructured by Hexi Chinese however Hexi Chinese is just some very generic kind of north western Chinese even if the strange thing is that is this the only thing we can know anything about Hexi Middle Chinese is by Tangud transcriptions and if we just read the if we just listen to Tangud's speaking very heavily Tangud accent Chinese it just sounds like normal Chinese without any Tangud accent what does that mean it means that Tangud is so much reshaped by Chinese so that the phonological systems are almost completely compatible with that kind of thing for example in Hamdou where you have melodic languages with phonologies so those people if they speak Chinese they just speak it like normal Chinese etc however we also know that for example for the compression for all kinds of compression the dialect the Tibetan dialect used to transcribe the Tangud and have shared the same sound changes so I think we have a typical situation which I call the the Ottoman Balkan situation you have the Balkan Shafut where Balkan languages coexist in some kind of in some kind of Paxpanic and they influence each other and they are all influenced by Turkish it's barely influenced by by Balkan languages and I think we have the same thing so Tangud and local Tibetan influence each other they are all influenced by Chinese but Chinese is not influenced is not very much influenced by Tangud and that's all I have to say thank you very much