 a peak at the advent of Tibetan in world history. So starts from the Yarlung Valley. That's where the Tibetan empire sort of spread from. Seems to have replaced earlier languages, including maybe most famously Zhang Zhong, which was spoken in what's now Western Tibet. And then there are two languages, Tibetan-Burman languages preserved in fragments in the library cave at Dunhuang. That also, let's say, show that more languages were spoken on the Tibetan plateau in those days. The alphabet was invented in 650, that's pretty clear. The oldest currently available text is the Zhol inscription, which is a pillar that sits outside the Patala Palace. Although that's not its original location. It was moved there by the Fit Dalai Lama. And it's from sometime after 763, not long after 763. So we have a sort of gap of about 100 years. Okay, and types of old Tibetan texts. We have pillar inscriptions, religious inscriptions on rocks and graffiti from Ladakh and adjoining areas and a couple of bell inscriptions, which are again from central Tibet. And wood slips and paper documents from the fort in Central Asia, of the fort of Miran in Central Asia. Okay, just a word on Tibetan dialects. There's a tendency for the dialects in the middle of Tibet to have developed tone and simplified their syllable structure and then for dialects on the outskirts to lactone and have complex syllable structure. And that's normal, right? This is the so-called neo-linguists in the 20s and 30s really emphasize that more peripheral forms are more archaic, geographically peripheral forms. There are various attempts to subgroup Tibetan dialects. None of them particularly convincing. So I'm not gonna say anything about them. But one point is old Tibetan. This is very important to me. It's a little bit controversial. Old Tibetan is older than common Tibetan. And what do I mean by common Tibetan? I mean, if you reconstructed Tibetan based on all the dialects. Well, how do I know this? Because we know from history when the Tibetans took over different pieces of the Tibetan Plateau and wider areas. And we know when they invented writing and they invented writing and 650 basically write at the beginning of this political project. So the form, the sort of state of language pointed to in the orthography is older by at least a few decades of older than the presence of Tibetan speakers in most areas. So that means that not to say it's useless to reconstruct common Tibetan on the basis of the dialects but it's not going to be very useful. It's like Latin where written Latin preserved in theological sources is more archaic than proto-romance, okay. So look at old Tibetan phonology. These are the consonant letters and the vowel letters. Kind of basic, normal, Indic sort of system. All of the letters, but this one, the 23rd are uncontroversial. I think that this 23rd one represents a voiced velar fricative. And a lot of people think that actually but a lot of people don't think it. And there's a lot of ink has been spilled on this question. And then just to point out that the letter W originally only occurred as a medial. Okay, so now moving to phonology. The script distinguishes voiceless aspirates and plain voiceless obstrants. So let's say P and pH, but the distinction was not originally phonemic. It sort of starts to become phonemic but it wasn't originally. Also, there are sort of special letters for certain kinds of parallelization. So there's cha for, you know, parallelized dental and so on. But generally speaking, parallelization is written with its own letter. So for instance, bia, you have a b letter and a y letter. But for cha, you have one letter. You don't have a t followed by y. And the voiceless L and the voiceless R are spelled differently. So in one case it's L-H, in one case it's H-R. So these are the consonant phonemes. So I sort of reorganized the consonant letters, gotten rid of these non-phonemic distinctions and these are the consonant phonemes. Look at phonotactics. Minimal syllable is a C followed by a V and a maximal syllable is actually four consonants followed by a vowel followed by two more consonants. But of those four consonants, if there are four, the first one has to be a B and the fourth one has to be a glide. Voicing is distinguished only in the let's say pre-glide position or the pre-vocalic position if there's no glide. So if you see something like B-S-G-R, then you say the whole thing was voiced because of the G. So that, so B-S-G-R is B-S-G-R, whereas B-S-K-R would be P-S-G-R. Okay, some changes from Old Tibetan to Classical Tibetan. Aspiration emerges as distinct but with a small functional load. There's this change of S-T-S to S. So let's only look at the third one. To listen is G-San in written Tibetan and is G-San in Old Tibetan. And a sort of, I don't know what to call it, a sporadic use of this final, what I think is a Vila Frigative in Old Tibetan disappears in written Tibetan, which is say the Old Tibetan forms on the left are one option in Old Tibetan. It's not always spelled that way, but on the right, that's always how the word in question is spelled in written Tibetan. But this character continues into written Tibetan in words with a syllable structure C-C-V because it's helping you identify where the vowel is. So this is, I mean, maybe there's no point in belaboring it, but this is giving you a sort of peak at the controversy surrounding this letter, which is a lot of people think it had in the final position a purely orthographic function in terms of indicating where the vowel is in the syllable. And I think that may well be an accurate description of what goes on in written Tibetan, but the forms on the left show you that it's not what happened in Old Tibetan. So I think it actually was a real phoning. This is something like the Shtambam of the Bodish family where again, I've taken it very, a very flat analysis because it's never been proven that a Bhumthang, Kirtop, Munpa, Zala, and Dakpa have anything in common in terms of a shared innovation. Those languages are all spoken in Bhutan and are referred to as the East Bodish languages. So I will be referring to them as the East Bodish languages, but I don't think there's any reason so far to posit an East Bodish subgroup. Okay, so now we're gonna start in on the historical phonology working our way backwards. First, Chung's law, which is the assimilation of B to ma before nasals. So if we look at a verb like to kill, which has a root satt, the past is psatt. So you form the past in this case at least by putting a B in front of something, but if the verb root starts with an N like in the verb suppress, then this B changes into an M. So the past tense of suppress is mnand. And similarly, the past tense of listen is mnyand. So quite normal sort of assimilation, but it makes the verb morphology or the verbal system look quite complicated because you have, oh, look, past can be signaled with a B or it can be signaled with an M. Okay, now Koblin's law, which is, I don't know, sounds a little bit funny, stated, well, I'll just say it. It's that you lose prefixes when you need to lose the prefixes. That is to say only certain things are phonotactically allowed in Old Tibetan. If you had something that was sort of morphologically motivated that would have violated that, you just dropped the prefix. So we'll look at some, not all, but some of these specifics. Let's start with the past formation. We already mentioned that you form a past verb by putting a B prefix on it. So let's say you have a verb like to do with the root bia, then you would form the past as bbia, yeah? But that's phonotactically impossible, so we lose the B. And similarly for sweep, we want the past to be something like bpyaks, and then we can't do that for phonotactic reasons, so we lose the B. It's, let's say a matter of taste to some extent, whether you want to analyze this as a synchronic thing, which is that sort of morphemes don't appear where they would be phonotactically impossible or a historical thing where you say that, like maybe there was a vowel there. So at one point in the past, you actually said bbias, and then the vowels disappeared, and then that's when it became phonotactically impossible, which is why it was dropped. I think both of those are options, and just in terms of what method we're using here, it's internal reconstruction, right? We're looking at oddities in the verbal system and trying to make them not odd, and that's how Koblin wrote a very important article in 1979 on the Tibetan verbal system where this is most of the work he was doing. And then similarly, let's say, I'm simplifying things, but in intransitive verbs, maybe not just in intransitive verbs, but there's a present prefix, which is this prefix, and it's phonotactically impossible in some cases as well, so it's dropped in Koblin's analysis, okay? Now I'm going to disagree with one of his proposals, and I think this turns out to be important, where he looked at, let's say, the verb kill again and said, oh, look, it has an o-oblout in the present and it has a g-prefix in the present. Well, maybe the g-prefix triggered the o-oblout, and maybe in a verb like to speak, we can say what happened was you had the root z'la, you added the prefix g' like you didn't kill, then the prefix g' triggered the oblout from o-to-o, so now you have g'z'la, and then because of Koblin's law, the g was lost because it's phonotactically not permitted, and then your left was z'la. So this is his analysis of the origin of the o-oblout, putting it another way, in those cases where you can have both a g-prefix and an o vowel, you have them, and in those places where you can only have an o but not a g, then you only have the o, and he sees that as a change and explains oblout as secondary. But if you look philologically, there are six examples of verbs where there is no g, although there could have been. In the present, but there is still an o, and there's also one example of where there is a g in the present that's not part of the root, and you can tell that because of the imperative, which is the shock at the end, right? It's not that the g is part of the root and it's kept across the paradigm. No, the g is the g that marks the present, but we don't have the o-oblout. So, Koblin's idea of the g causes o-oblout is wrong. I think that probably, it's not wrong totally, which is to say, at some point, there were enough examples of a correlation between g and o that that may have analogically spread. So, his analysis is sort of pointing to a moment in the productive history of morphology in Tibetan, but it's not the original situation. And then the point is, this means we can't just miss o-oblout in the Tibetan verb, and that's very important to me, because all historical linguists who work on Tibetan and all of them that work on Chinese for that example, ignore all things that look like o-oblout, and I don't think that's helpful. Okay, oops. So, now on to the next slide. So, now on to just more sound changes. So, this is Dempsey's law, the merger of e and i before dealers. So, here are examples with e. The e is preserved in Chinese, right? We saw a very similar change in the history of Burmese. So, look at number one, you have tech in Chinese, you have chick in Tibetan. And examples with e, you look at the word for joint, you have, you know, sick in Chinese and sick in Tibetan. Just to point out that this is not such a crazy change, it happened in Latin as you see if you look at moisen, which is tingo in Latin and tango in Greek. Now, we want to look at these changes and understand how old are they? Are they characteristic of the Bodish branch altogether or just Tibetan? And in the case of Dempsey's law, it's not clear because Kurtop has e in all cases. So, let's say it seems to have also undergone a merger, whether it's the same merger at the same time is a little unclear. So, here's the evidence where when Chinese has an e, Kurtop also has an e. For example, in name, where you have meng and meng. Oh, but, yeah, oh, but further down on the same slide, if you look at harvest year, Chinese has ing, but Kurtop has meng. So, yeah, so Kurtop also merged, but not maybe to the same vowel as Tibetan. Now, Dempsey's law would predict that there are no syllables that have the rhyme ng in Tibetan. Why? Because they should have all changed to ing. But if you look around, you find some. So, here's sang and it means purify or clean. And I've done some work to try and kind of explain the etymologies of these, although I, you know, there's still a lot that I haven't explained. But here is my attempt with this one. You start with a model. This is the verb to do. So present, past, future, imperative, bied, bias, bia, bios. We have that in mind as a normal verb, right? And we solve for x, oops, I should have had a line right there in the following equation. Sorry that this didn't display better. Bied is to bias, sorry, other way around. Bias, the past, bias is to bied as besang, which is the past of clean purify, is to what? And we solve this and we get sang. And that's my analysis of why you have this syllable sang, is it's an analogical restoration based on forming new presence to inherited pasts. And as evidence that helps me, you know, make this claim, we have an etymologically related verb that's an extension of the inherited present, which is g'sing. So to tell the story forwards in time, I think there was a time when you had a present g'sing and the past besangs and then g'sing changed into g'sing according to Dempsey's law. And then that verb was so funny looking, it was too much even for the Tibetans. So they specialized g'sing in a slight different meaning, where it continued to have a life on its own, and they restored a new present that sort of violates Dempsey's law sang. I hope that was, I don't know, clear. This is my attempt in a sense to show you, I think how to use historical phonology and analogy as your two tools, right? You sort of make a theory based on historical phonology. You say, well, I'm a neogromarian, so I believe it's an exceptionless change. So that means I have to hunt down every single exception. And then one of the tools I have for explaining those exceptions is analogy. And the most classic form of analogy, although not the only one, is this four-part analogy where you basically say someone forgot how to say, the present of clean in this case and invented it based on existing structures elsewhere in the language, okay? That was enough of Dempsey's law. We move on to Benedict's law, which is the palatalization of le, of liya, let's say, to ja. Field, for instance, where Chinese shows, we started with something like ling, sweet or tasty, ground, bow, although it's a little strange that Chinese uses the word arrow for bow. And there's Tibetan internal evidence for this as well. So you look at this, so this is Gong Huangcheng proposed that Tibetan had an honorific infix y. So you have something like scum means to be dry and you can change it to a skym, the honorific of to be thirsty. And there are pairs and you see them on the screen between L and ja. So his analysis is you have something like lock, sorry, I should pronounce the S at the final, locks means side and you want to make it fancy sounding, so you add an L and a G apparently. So you get ge yox, but then the sound change makes the liya into a ja. So you have ge jox, okay. So that's internal evidence for this Benedict's law. But an important point is we still have more, I won't get into it, but we need the little y there as a conditioning environment. I'm using the little y there to say, this is the conditioning environment for the palalization. Yeah, okay. And I think there's limited evidence, but real evidence that something similar happened with R. So the best example is definitely this word for day because Burmese has a ryak, r-y-a-k and Tibetan has jak. So then it looks like the ry, like the ly changed into ja. Okay. But what do we do about these words that have a la? I'm actually skipping a logical step. What do we do about words that start with la in a seemingly palatalized environment in general? That's the question. Well, I have an explanation for all of them before ik. So we look at things like island, ling, flute, ling, hunt, ling. And my suggestion is that Benedict's law happened before Dempsey's law. So if we look at field, which turns into zheng and hunt, which is ling, we can do something like say that originally field was ling and hunt was ling. And then Benedict's law came along and changed ling into zheng. And at that point we had ling and ling. And then Dempsey's law changed ling into ling. I'm worried that that was so too much too quickly. But basically the ii vowel is the conditioning or one of the conditioning environments for Benedict's law, right? So in the case of words attested with a la initial that end in ing, we can speculate that at the time of the application of Benedict's law, the rhyme was not yet ing, it was still ing. And we can do that by proposing that, you know, that Dempsey's law had not yet happened when Benedict's law happened. And I think that's, I don't know, one of the things that is one of my sort of hobbies is to work out relative chronology of sound changes by tricks like this. I just want to point out that tasty is a problem because the Tibetan vowel doesn't match the Chinese vowel. And if you started with lem, it should have been preserved in Tibetan. So there's something funny going on here. The Chinese character is quite late attested. I sort of want to believe that they're cognate, but there's a problem here. So did Benedict's law happen in East Bodish? And the answer is no. You know, just look at the last word on here for where Kurtop has bleh and to Ben-Naz-bji. So Kurtop did not parallelize the L. Now, Benedict's law doesn't get us all the palalizations we need. If you look here, we have kick for to tie, fasten or suffocate. It's not kick, it's kick. And two, we have, let's say, knee in Chinese, something like knick as the proto-Burmish form. But in Tibetan, we have gynese with a nia, not with a na. So why all this palalization? The easy answer is Tibetan parallelized always before the vowel it, but that doesn't quite work. We have a few examples where there's not palalization before it, and we also have examples where there's palalization before other vowels. So I throw up my hands and just say, there are some things we can figure out around Tibetan palalization and there are some things we can't at the moment and it needs more study. Did East Bodish undergo this other form of palalization? And the answer is it did not. So let's look at this one tree, because we've seen tree in a lot of cases, shing in Tibetan, but sang in curtop. So no palalization. Yeah, and then because we can't explain the origin of palalization, what I do in my reconstructions is I put this little y in there so that I write rather than shing, I write se ying. Okay, so that was it for palalization. Now we go on to Khan-Radhi's law, which is dental excrescence after this rhe. This is again, we're back to the verbal system. So you see we have a root like so. So so means to nourish and it's present is khut so. Well, the analysis here is the h is the present prefix, but you get an appendices where you don't just say khut so, you get khut so. That's a way to make again, the verbal system a little simpler. And then here's an example before voiced consonants and it also happens before r. So rhe is the stem for to write, but the present, sorry, rhe is the root for to write. It comes up in something like a rhe mo picture, but the present is khud rhe, where there's been a dental inserted, right? This also happens before laterals, but the situation for laterals is more complicated. And if you're not interested, you can just sort of daydream for two minutes, but so we start with rhe ol, so the 23rd letter before l, we get the dental appendices exactly like we did in the previous cases, but then there's a metathesis and then Koblin's law means we drop the initial rhe. So you end up with a very funny looking sound change, but it's not so funny looking when you look at it in steps where sort of gamma l changes to ld. But there are examples that really make it pretty clear. I think so. Choo, the root is ld, ld, and the present is ldad. So it works in terms of what's going on morphologically is you're just applying this present prefix to the root. And it tells you that that morphology was productive before all these sound changes happened. And a similar, or at least a component of this change, which is the metathesis of dl to ld also took place in the history of Spanish. Okay, now again, just looking at the interaction between these changes, how do Conrad's law and Benedict's law interact? Well, let's look at flea. In the word for flea, the l did not parallelize. So that means that the l was not in right before the eval when Benedict's law happened. So I think it went something like this. You had hallibah, then you got the dental appendices. Hallibah, then you got the, what the metathesis. And then it's the d that parallelizes, it's not the l. So you can tell that the excretions of Conrad's law occurred before metathesis. And so Conrad's law must precede Benedict's law. And that's one reason you can see, you know, that I'm going backwards in time. I talked about Benedict's law first. Now I've talked about Conrad's law and now I've shown that Conrad's law has to be before Benedict's law. Okay, now Bodman's law, this is a fun one. M-L changes to M-D, which gets you some really nice sign of Tibetan cognates, in particular arrow where all Burmese has mla, and Tibetan has mda. So we can just say that Tibetan originally had mla. Yeah. Also for fathom, that means the span of distance, right? One fathom, it doesn't mean the verb to think. So you have lam in Burmese, so lam in Chinese. And we probably have malom as the prehistoric form in Tibetan. Okay. There's also Tibetan internal evidence for Bodman's law as we see in this ensemble of words having to do with blindness. So we have madongs, ladong, and long. So that makes it clear that it's quite reasonable to think that there's a lateral at play in the root that it has to do with blindness. And you get different outcomes depending on the different prefixes. Okay, how about Bodman's law in East Bodish? Well, it didn't happen. So far, East Bodish has not participated in any of these sound changes, I just want to point out. Yeah, so the mumpa word for arrow is mla, just like in Burmese. Okay, so now I look at Bodman's law and Konradie's law. So if Konradie's law occurred after m, Bodman's law can be seen as a sub case of Konradie's law. It's dental appendices, right? Because of the L, and then you could have metathesis and then you could have Koblen's law that deletes the L. But I don't like this explanation. And the reason why is because then we have, in some cases, Koblen's law is deleting the re, the prefix. But in other cases, it's deleting the L. Yeah, so I don't like that. So I would rather choose to analyze Bodman's law and Konradie's law as unrelated, but I just wanted to say, some people have done it this way and you might be tempted to do it this way. Okay, now a exciting little change, which is that Uba changes into wa. And this actually allows us to get rid of all medial Ws in old Tibetan. So you notice that all old Tibetan words that have a medial W have it before the vowel a without anything following. So this, I mean, this isn't good motivation for this proposal, but it is a consequence of this proposal, right? If you changed Uba into wa, then you would only get wa in this circumstance. We should say open syllables where the vowel is a, okay. But also this proposal explains the alternations that we see between a root that has U and the medial W form. So let's just look at horn. There are two words for horn in Tibetan. Rhu and rwa, it tends to be that the shorter form happens in compound. And this is also true of, let's say, like there are two words for hand. One is lak in compound and one is lak pa as a free word. So this would make this wa have exactly the same function as this pa or pa, nominalizing suffix. So I think it's a brilliant analysis that Guillaume Jacques and Nishida, rather, have come up with. Okay, now on to just another change. Continue going backwards in time. Some of the Ys in Tibetan seem to come from W. So for instance, the verb to be is yin in Tibetan and it's wen in Kirtop. I think that's the best example, but let's look at dog as well. We have ki in Tibetan and you have qui in Kirtop. So which Ws changed into Ys? Was it before the vowel i? Well, then you don't get the weed example, but the weed example maybe isn't such a great example to begin with. It's not totally clear to me. But you do also get, let's say a yy correspondence between Tibetan and Kirtop, which just proves that it's a merger, right? So Kirtop distinguishes ya and wa in these cases and Tibetan doesn't. So I'll just sum up and say, there's something that I think is kind of cool here, which is here, this sound change gets rid of all the medial Ws in all Tibetan, but this sound change gives us some initial Ws that we didn't have. So our reconstructed language is starting to have a very different sort of phonotactics than all Tibetan. Okay, Laufer's law is that basically labial velars and labial uvelars induce an O vowel. So this explains a lot of cognates to Chinese and I will point you to the third one. So you have a grot in Tibetan and kwa in Chinese for stomach. Okay, the R is a problem, but otherwise they are pretty similar looking words for stomach and this proposal helps bring them closer together. It also I think indirectly helps confirm the reconstruction of labial velars and labial uvelars in all Chinese, which I can put another way as it would be reason enough to reconstruct labial velars and labial uvelars in the history of Chinese just to make correspondences externally better, but we also have a lot of internal reason for doing it. Okay, the word for bear is a problem though, where we have an O in it in Tibetan, but there's no O in Kirtop. There is an O in Munpa, kind of confusing. No O in Chinese, no O in Burmese. So it seems like the root, if you like, is something like a Wham, but it should have a G at the beginning in Tibetan if it were to work perfectly, right? Because you would say, oh, it's this labial uvular in Chinese that changes to a W in Burmese. That's totally regular. The schwa changes to an A. So this predicts a form Gom in Tibetan and you don't get Gom, you get Dom. It's a problem, but the same correspondence to me seems to come up in the word for wing, although I think most other investigators don't think these two words are cognate. Okay, there are other origins of O, which is to say Tibetan has a lot of O's. Tibetan has a lot of O's and they have a lot of origins. So A W is one which you see that in Chinese. It's A W in Tibetan, it's O, okay. So now we've gotten to Proto Bodhis. And I would say in approximate order of changes, we have Lauffer's law, W goes to Y, Uva goes to W, Bodman's law, Konrati's law, Benedict's law, Dempsey's law, Koblin's law, Chong's law. And then conclusions about what Proto Bodhis looked like. It had a wider distribution of L. Tibetan W is always secondary, but there were some initial W's in Proto Bodhis, but there weren't any. This is an oversimplification based on Nancy Kaplow's dissertation that Gongxun has some issues with and I think he's probably right, but he hasn't published his, he has published a clear statement. So we'll just go with Nancy Kaplow's presentation. There's the ba that is on verbs and there's the ba that is on nouns and both of them are traditionally seen as some kind of nominalization, but in the case of verbs, it clearly is a nominalization because it changes something from being a verb to a verbal noun. Whereas in nouns, it just sort of helps the noun be used as a free word, right? Like Lak means hand, Lakpa means hand. It doesn't, there's no nominalization going on there, but there is a suffix. So when you teach Tibetan, you treat these as basically the same kind of suffix, but their tonal effects in Tibetan dialects are different. So if you were sloppy, you would say, oh, well, you know, ba behaves differently in verbs than it does in nouns. Whereas if you're neogrammarying, what you would say is that those tonal differences that have something to do with the accent system in old Tibetan were there in the proto-language. Where, yeah, in the, what, I mean, were there in common Tibetan? And I don't have the information in front of me, but one clear example also is na, which can mean in or if in Tibetan, yeah? So it means in on nouns and it means if on verbs. Gongshun actually, this is part of his discussion as shown that they, across all Tibetan dialects, they have different tonal effects. So you can show that they're two different morphemes. So anyhow, that's why I put that accent there, was to say, you know, it's this ba, not the other one. At the risk of sounding like a hypocrite, you can understand the change as Uba did not change into wa in verbs. That's not neogrammarying, but it's a helpful mnemonic, yeah? What you could say if you're a neogrammarian is that either it's a different ba suffix, which is what I was saying, or you could say that it did change. So a verb like to steal was originally rkuba, and then it changed it to rkwa, but then rkwa looked so aberrant in the verbal system overall, and there was a ton of analogical pressure being exerted on it that rkuba was restored and rkwa was lost without a trace. That's, you know, which is to say, I have two stories I can tell, and both of them are neogrammarian. We don't have evidence independent of the kind of stuff that Nancy Kaplow and Xun Gong are talking about. And I would say that it manifests in the tonal systems of the central dialects, but is related to accent stuff, word level accent, in the non-tonal languages. So it was probably some kind of, let's say, barely phonemic word accent system in the level of kind of all Tibetan synchronically. Now we're going to head further back in time from Proto-Bodish to Proto-Trans-Himalayan. Liefang Kuei's law, this is one of the real exciting laws in all of the sign of Tibetan family, if you ask me, which is that a rya changes to argya. And one of the, let's say the best example, maybe, is the word for hundred where Tibetan has brgya and Burmese has a rya. So Liefang Kuei said, well, yeah, maybe the g is empathetic. There are typological parallels for this too that are in my book. Those of you, so aside to those of you who have noticed it, am I proposing that rya becomes two different things? One is rgya and one is zha. The answer is sort of, but I think we need to index the difference. So I discussed that in my book, yeah. Okay, and now about the timing of Liefang Kuei's law, and this is a real mystery, which is why I presented to you. On the one hand, there's evidence that it was sort of just happening as Tibetan enters history. So Udyana, which in some kind of new Indo-Aryan would have been the, would have been an R, so something like Uryana becomes orgyan in Tibetan. So it's a lone word where they've inserted a G. That means it's a new change. And then there's this word for a divination board that in all Tibetan, the orthography is unstable, even in a single document between rgyaks and rgyaks. So that really makes it seem like we can see this sound change happening before our eyes. But bizarrely, other Bodhich languages seem to have undergone the change. So Kurtup word for eight is jat, and in Munpa, the word for eight is get. So the G is definitely there. Now you could try and get around this. You could say that the East Bodhich languages borrowed their numeral system from Tibetan. Yeah, it could be, but that seemed a little bit abhoc. So it's a real contradiction. There's evidence that Lifang Khoi's law is really new, and there's evidence that Lifang Khoi's law is really old, and I don't know how to solve that problem. Now Sakya Pandita's law, which is that G changed to D before graves and D changed to G before acutes. Now, as a synchronic fact, Tibetan D and G as pre-initials are in complementary distribution. So you could reconstruct there as being only one source. But Sakya Pandita's law, and this is kind of Jacques's interpretation of it, is that D and G were originally separate and have collapsed. And so the best evidence for this is the word for ant, where, so Jacques says there's a G prefix in animal names, and he says, oh, look, there's ant, there's leopard, there's eagle. You all know that I have a certain amount of skepticism around animal prefixes, but fine, I'm okay with this. And I'll just point out actually that the word ant works extremely nicely in terms of you have a uvular in Kyaurong, and then you have an R initial, which could come from a uvular in Burmese. The point here is that in Tibetan, you could have a D there. You could, if the word for ant were drog, you could have written drog, but it wasn't. The word was grog. So then maybe in leopard and eagle, the G is original, even though you couldn't have a D there. And then he similarly thinks that in body parts, there's a D prefix. So in I and in hip is evidence. So I think that's Sakyapandita's law. And now Houghton's law. So Houghton's law is another kind of palalization. So this is our third kind of palalization, but this one's really old. And it only affects velar nasals, and it doesn't affect all velar nasals. So again, the conditioning environment for the palalization is hard to pinpoint, but how do I know it's so old? It's because in this case, the East Bodish languages share it. So East Bodish languages don't participate in Benedict's law and they don't participate in the overall palalization in Tibetan, but they do participate in Houghton's law. So the word for fish, which, you know, remember in Burmese and Chinese is nga, basically. Well, in Kurtop and in Tibetan, it's nya. So yeah, so now we have our first isogloss that proves that Bodish is a family, if you like, okay. But gums, which I have treated as an example of Houghton's law, because it starts with velar nasal in Chinese and Burmese, it has a dental nasal in Kurtop, which is to say, maybe I'm wrong to treat gums as an instance of Houghton's law. Maybe it's actually this other later palalization, okay. And now my all-time favorite sound change that other people don't believe, Guillaume Jacques in particular, is that I think us changes into us in the history of Tibetan. And the evidence for this is in two weird verb conjugations. So it has the conjugation za, zos, bza, zo, and chu is cha, chos, bcha, cho, okay. So the o-abla in the imperative is totally regular. Don't worry about it, but having an o-abla in the past, it's bizarre, it only occurs in these two words. And these two words, I would point out, I have certain phonetic things in common. They're open syllable root in a. So my proposal is that us change into us. Now that means, oh yes, first, this proposal does a lot of work for me, which I'm very happy about, where, okay, you say, well, Nathan, why didn't us change to us in the past tense of to look at? You have leta, letas, leta, letos has to look at. And I would say it did. And that's why we have a verb letos, which means to look to or to attend to. And then what happened was, in that case, because the verb was less frequent, if you like, that irregular past went off and formed its own new verb, and then letas was analogically restored, and I would tell the same story about the other two verbs here. So I think this as to us is already doing a lot of work for us, yeah? And then you say, well, you know, what about all these other verbs, you know? So bie, bias, bia, bios. Why isn't it bie, bios, bia, bios? I would say analogy, right? There's gonna be a huge amount of analogical pressure to undo the sound change. And we even see that happening in historic times with the verb to eat, where bazaas replaces zos. And if you're skeptical, you say, okay, fine. You say there was an exceptional sound change that changed all arses into os, but it's the only direct evidence is in two verbs and indirect evidence in another three verbs. It just seems awfully speculative, Nathan. But then I'll point out that curto, which usually has ar as the cognate of ar, in the words do, borrow, eat, and devour has ooh. So this suggests to me that this os change happened on the way from Transmolayan to Proto-Bodish and it's shared by Proto, by Kurtop Antevan. And that's why you get it in exactly these words. So I think this is beautiful. I think it's really, really nice. So the only, well, so Bettina Decisor doesn't like it and you can read her reply to my article, but I have a little trouble following what she's trying to say. Guillaume Jacques doesn't like it because he thinks this oto or oblate is related to the agreement system, which he thinks it's sort of evidence of the agreement system and his solution works in a sense. It can explain why it's retained in Kurtop Antevan, whereas as far as I can tell, Sizler's explanation will not get you the Kurtop forms. But I don't think, like, I like Guillaume, want to see agreement morphology in the history of these languages, but I just think Occam's razor says, you know, look, this is a problem we were able to solve with exceptionalist phonology and analogy. So let's just leave it there. That's my feeling. Okay, now on to Schieffner's law, which is that Z changes to Z. Schieffner, and I think this is one of the earliest sound laws that has been proposed in terms of the history of the discipline. He said, well, look at the future of this verb. So we have Zook in the present, but Zook in the past, Zooks in the imperative, and we have Zook in the future. It would be nice if instead it were Ge Zooks. Sorry, Ge Zook. So that's his internal reconstruction. And there's other sort of internal etymological evidence for it. So we look at Zong, which means merchandise. It looks like it has some relationship to the verb to sell, which is Zong. But it would look even more like that if it started with a D. The comparative evidence for Schieffner's law, so you see it there, you know, with Eat, with Bridge, with Coral, it's really pretty impeccable, I think, that Zook changed to Zook in certain phonotactic positions in Tibetan. Now let's just look at this verb to see sort of Tibetan historical phonology as told by the story of one verb. So the current conjugation is Zin, Zong, Ge Zong, Zong. So I think we started with something like Ge Zong, Ge Zong, Ge Zong, Ge Zong, Ge Zong. And then in with Schieffner's law, we get Ge Zong, Ge Zong, Ge Zong, Ge Zong. Then I didn't talk about this change, but in the present, you get this U-T-E change, which is sort of somehow triggered by the final D. And then you get an assimilation of the N to the D. And then you get Sakyapandita's law. And that's what changes the D prefix in the future to a G prefix. And then you get Konradi's law, where we restore the D, right? Because you see that we changed the Z to a Z from Schieffner's law, but then Konradi's law, we stick the D back in, the dental excrescence after the prefix of a. Okay, now how about Schieffner's law in East Bodish? Well, it's there. You see that we have a Z in these words in Tibetan and we have a Z in these words in Kurtopin Monpov, which is to say, I think we're starting to build up some pretty good evidence that there really is such a node as Bodish. They share Schieffner's law, Houghton's law and this very particular Astaos change. Okay, now Simon's law, which is that mra changes to bra. This is a very helpful sound change in improving sign of Tibetan etymology. And let's look at the second and third example. The word for a female yak in Tibetan is, well, brimo, brimo. And in Chinese, it's mru, probably a lone word from Tibetan. Yeah, but an early one before Simon's law. So, yeah, so there's some example, that was an example. And, well, let's not look at nomad because it's also kind of too Tibetan specific, but rather be where we have a word for fly mrang in Chinese and this change allows us to reconstruct smrang in Tibetan. So those start to, so Simon's law does a lot of work for us in terms of improving sign of Tibetan comparisons. And how about Simon's law in East Bodish? Yeah, I really wanted to be there, but I just, the East Bodish languages are extremely poorly documented. I've gotten all these cognates from just sort of articles about totally different subjects. There's no good, you know, Kirtop dictionary or Munpa dictionary or big fat grammars. So, you know, the information's just not there, sorry. Now, there are some exceptions to Simon's law, which is to say there are words that have m followed by r in them, which you don't want, right? If mr changed to br, then you don't want to see mr in Tibetan. But there are these words. None of them seem to have good Simon Tibetan cognates. I don't know how to explain them. And now, a proposal of Jacques, which is that Rlya changed to Rya. This is quite similar to other stuff we've seen in Tibetan, let's say, to, it's a kind of, let's say it's particularly similar to Bodman's law, right? Because we're getting a fortition of a ladder. And here's some evidence of it. The only reason I put it sort of this far back in time, if you like, is because I think that the evidence is not super strong. Okay, now the merger of a and a, this should be pretty familiar by now. You know, Chinese has a and Tibetan has a in words like armor and shell compared to each other or hill and hill or five. But there are other cases where Tibetan has a and Chinese has a schwa, like weave we've already talked about in the history of Burmese. Tibetan merges a and a before dentals, including R and L. We can say before acutes, if you like. A change should also happen in Burmese, right? So here are examples with a, so Tibetan has, Tibetan and Burmese have a in the word for eight, but Chinese has E. Tibetan Burmese have a in the word for cut, but Chinese has E. Some people instead want to reconstruct this as a ya in sign of Tibetan that changes to a in all Chinese. I don't like that proposal as much. I think because it complicates the phonotactics of the proto language, but you know, maybe it's right. Okay, and then here are examples with a, okay. There are two words where Tibetans sort of, let's say has kept the E where we don't expect it to. One is to increase and one is sleep. Now into increase, I could tell you a story about analogy, although it would end up being a complicated story. But basically that, you know, one way or another, we analyze the verb as a present tense verb and so we stick the E back in it, but that's not going to work for hail. So I don't know how to explain these. Okay. Now Piero since Taros's law and we're getting sort of far enough back in time that you're seeing, you know, laws you've seen discussed for other languages. Tibetan merges uvulars and dealers. These examples are uvulars where Chinese preserve the uvular and Burmese drops it. And these examples are dealers. Where all three languages have dealers. Lovely, huh? There's one exception where we have this, this dealer voice dealer fricative instead of a G in Tibetan. I don't know why, but I think it's probably something phonotactic because a word gong, just G-O-N-G, gong would look really odd to me in Tibetan. And I had trouble putting my finger on why it looks odd, but it seems odd. So I think this might have a kind of later Tibetan internal explanation, but I don't know. And then also we've mentioned bear before, but now it's coming up in the context of uvulars, which is we have a reason to reconstruct the uvular in Chinese and that's consistent with what happens in Burmese, but we have a dental in Tibetan for some reason. Now there is evidence in other places, including Tonggut, which I didn't include here, but Situ Gyarong has the wong for bear. So, you know, so one thing you might say, and I think this is probably what Guillaume says, is that actually the Chinese word didn't actually start with the uvular here, back strings are wrong. It started with an initial w. I would rather not complicate, again, the phonotactics of the proto-language that much, just for this one example, but maybe that's the answer. And then you could basically see the Tibetan form as going back to something like the Situ Gyarong form with the syllables being smashed together. Okay. And then here's a correspondence we've seen before, which is u and u, we, you know, u in Tibetan, u in Chinese, we reconstruct as u, o in Tibetan, o in Chinese, we reconstruct as o, but u in Tibetan, oh no, wait, it's just a mistake on the slide. I have it right in the title. u in Tibetan and o in Chinese, I reconstruct as schwa w. I feel like this is a very tentative solution, but its advantages, as mentioned on a different day, are that it fills a gap in all Chinese, and it's only called for in those situations where you also have things like o and u in Chinese. Problems are that the schwa w examples outnumber the o examples, which seems unlikely to me, and that if o and u merge to o in Tibetan, we would expect schwa w to merge, to o in Tibetan where it doesn't, it merges into u. So we've seen this argument before. In any case, here are examples that can be reconstructed o, including after birth and uncle and nine and elbow, and here are examples that can be reconstructed o, sharp, relaxed, servant, and then here are the words that can be reconstructed as schwa w, so steel, bends, meat. Now on to our next change, Tibetan loses final y. We've also touched on this before. Here are some examples, so net or trap, something like ria, whereas you have rai in Chinese. And then sometimes, I don't even know if I formulate this observation correctly, but sometimes in these words where you would have expected the final y, Tibetan has an e vowel instead of an a vowel. And is this some kind of condition change? Because of the y, it's not clear to me. Maybe it's oblaut that's inherited from the proto language. It's not clear, it needs to be solved. And in two words, Tibetan has a d, which I guess we have to assume is some kind of suffix. It's not there in Chinese or Burmese. Okay, and then rl changes to l. I mean, this is a definitional thing, right? Where Chinese has an r, Burmese has nothing, and Tibetan has l, I reconstruct rl. So these are the examples we've seen them before. And there are ambiguities between rl and r, where Chinese has r, but Tibetan has both l and r. And particularly this word for neck, it needs philological work. Maybe it's a one-off, maybe it's borrowing inside of a dialect, or maybe it's old and interesting, needs work. And then origins of final h, or let's say final rl. And this is one reason why I think it's real and not just an orthographic nicety, is that it's cognate with k in Chinese. And as you know, and this is a solution I feel uncomfortable with, I reconstruct this as a k followed by a shuang, in words like a hundred and arrow and pass. And I think it's, this is my defense of my reconstruction, is that then you could have the sequence of changes where you have lenition and then apocapy leaving the r in Tibetan. And in Chinese, you could have basically apocapy before the lenition. Okay, there are some exceptions where Tibetan has this final rl and Chinese has a ya instead of a k. I don't know how to explain it. And then here's another exception where Chinese doesn't have anything, but Tibetan has a final a. And then also the k and q collapse as, it's written as a g, but it was pronounced as a k in Tibetan. So here are examples that we can reconstruct as with a final k. So night, for instance, or a 24-hour period in Tibetan cognate with night in Chinese, both something like a rak and then one both going back to something like tech. Okay, and then here are the examples with q where Tibetan has a final k and Chinese has a final gul stop. And then the last topic, and oh, I don't really have time for it, but I'll try and plow ahead is the age of voicing, alternation and oblate in Tibetan. Or basically, I just want to say that most people think these are late and I think they're early. So someone like Randy Lapola or Metsulin think that voicing alternation is due to segmental prefixes and that you have some very simple, like voiced is intransitive, voiceless is transitive. That is just not the picture in Tibetan and it hasn't been the picture. Well, I've written the date as 1593, it's actually 1953, which is still a long round ago since Ure pointed this out in 1953. We have three types of verbs. We have the A voiced, the B voiceless transitive and then we have a, sorry, I'll just say this again, I'm trying to rush, but I shouldn't. We have type A, which is a voiced intransitive. We have type V, which is a voicing alternating, voice alternating transitive. And then we have a voiceless intransitive as well. The C forms, I won't talk through the details, seem to be derived from the B forms. And then here are our examples, I'll just talk you through the first one. So you have growl voiced throughout the paradigm to be free, growl with voice alternation inside the paradigm to liberate and crawl with voiceless throughout the paradigm unravel. So the first one and the third one are intransitive and the transitive one is the voice alternating one. So just this picture has not been grasped by the Sino-Demedanists, if you like. They've seen a much more simplified picture and it has never been sufficiently addressed. I'll run through this and just say, that's my argument that voicing alternation is old in Tibetan and needs to be handled. It needs to be handled at the Sino-Tibetan level because no one has handled it at the Tibetan level, even though some people have thought they did. And similarly for Ablau, and I've sort of looked at this already a little bit that this G causes A to O, I don't believe it. And then there's also this idea that a D causes A to E. And I just want to point out that I think there's evidence for Ablau already back in Sino-Tibetan. So we have to know in Tibetan is Macan and then to see is Ken in Chinese. We have to see is Matong in Tibetan and it's Marang which probably comes from Matang in Burmese. We have to read originally to chant in Tibetan which has the very exciting paradigm clock, black, clock, clock and it's cognate in Chinese is lock, which looks like the present, right? So which is to say, I think there's some kind of Ablau going on at the Sino-Tibetan level. And that is something that I wasn't really able to address very well in my book other than to just point out that I think it's real and you don't even need Tibetan to show it. That's what, like I think the Ablau is most robust in Tibetan of these three languages but there is evidence just looking at Chinese and Burmese and even just within Chinese. So voicing alternation and Ablau are old and people need to take them seriously which they haven't so far. And then I won't talk you through this one because I'm already over time but it's just a summary of the changes that I've already discussed. And so trans-Tibetan from Tibetan perspective had distinct eulers and vealers, had a sixth vowel, had no vowels, had no voice fricatives. And from the Tibetan perspective, the glide ya or let's call it the feature of palization occurred after vealers, nasal, la and hurrah. Okay, that's it.