 Okay, so we're picking up where we left off yesterday, but I thought it would be useful to just give a slight orientation again. So here we go. We're looking at the sources and the methods of old Chinese reconstruction. And in terms of sources, we talked yesterday about the rhyme tables and the rhyme books. And then today we'll talk a little bit about early poetry and the structure of the Chinese script. And so that's what I'm gonna go into now. And then we'll talk about methods. Okay, so if you look at one of the odes in the book of odes, as if, you know, I don't know, as you will find it on the internet or in Song Dynasty print versions, and read it using the, you know, opinion pronunciation of Putonghua, you will be able to tell that it's intended to rhyme. Now, so if we look at here in the second stanza, you have Dua and Dua rhyming. And in the third stanza, you have Xie and Xie. So it's clearly intended to rhyme. But in the first stanza, you see that Xie and You are rhyming and that doesn't seem like a very good rhyme in Putonghua. So I think this is a good sort of point of departure in terms of saying, you know, despite everything that's happened in the last 3000 years, historical phonology is sufficiently regular that you can still tell these poems are intended to rhyme. But they don't necessarily, well, of course, no surprise, they don't rhyme everywhere you would expect them to. So a theory of Chinese historical phonology would improve these rhyme patterns. And then to some extent, it's a question of taste, how far to push that, you know? Sometimes maybe rhymes weren't perfect, even in the original. But anyhow, it's an important source of evidence for what things sounded like in the old days. It's a question of genre, basically. And in old Chinese, let's say, when we say old Chinese, we really are talking about the poems in the Book of Odes. And in the Book of Odes, there's a certain amount of variation in different parts of the Book of Odes. So some poems don't rhyme at all. But what you see here is probably the most frequent pattern, which is that, I mean, I'll say it in a way that you will find a paradoxical, but then I'll caveat it, that the end of the second and fourth line rhyme. And then you say, well, look, in this case, it's not the end, yeah? And the issue there is that there are certain words, certain morphemes that are considered to not be counted. And jer, which is what you see here, isn't counted. So I mean, like, why did they do it that way? Well, that's a question that's hard to answer. So the reason I chose this poem is actually because it's back through 1992's poem. And what's nice about it is you see that the first and the third lines repeat exactly. So we can just sort of totally not worry about it. I think that's why this seems like a good example to pick, because sometimes you get rhyme also in the, you know, between all the lines of a stanza or there can be quite complicated patterns. So actually it's, I'm not going to go into it in this course very much, but understanding what the rhyme scheme of particular poems are is also one of the kind of moving parts in Chinese historical phonology where you can imagine that, you know, in one proposed reconstruction scheme, in one poem, you'll get different rhymes than in another reconstruction scheme. So one of the things that people can argue about is exactly which rhyme patterns are intended. Oh, so here's another example. Let's, if we just limit our world to this poem, what I would say is we both agree that the fact that that side-side fo-yi rhymes with side-side fo-yi is not interesting. Yeah, it's not going to tell us anything about Chinese historical phonology. Well, similarly, the fact that sort of jr rhymes with itself is not interesting. So the fact that jr rhymes with law and the jr rhymes with xie can suggest to us that tsai and yo maybe should rhyme. And I think you should take it that way. It could be that that's incorrect, right? That's a hypothesis. And it's also, let's say, a hypothesis, they can't be proven in a sense because we can't ask the author of this poem, did you intend to rhyme this and this? But you can pose it as a hypothesis and then see whether these two characters rhyme in other poems, or you can see how the analysis of this poem in the two ways that we see as possible, which is assuming that these two words did rhyme, or assuming that they are not intended to run, can interact with other hypotheses. And now the structure of the Chinese script. So this comes a little bit to the nature of writing and its relationship to speech. But although they're very opaque, there are a lot of other hypotheses and so they're very opaque. There are indications of pronunciation in the Chinese script itself. Here we will, I will look at this Shesong series built on this character that you see in front of you. And what's a Shesong series? It's all of those characters that share a phonetic determiner to use Boltz's terminology, which I think is good. So we have the kind of origin point for the series. Yeah. And then we have other characters and you see these are the middle Chinese pronunciations. If you looked at them up in the Guanyun, right? And you see that they're all kind of similar. Yeah. They all start with labials and the rhymes are either ah or eh. Yeah. So there's clearly a, you know, given parameters of variation in terms of what characters that share a phonetic determiner have in common in terms of pronunciation and what they, you know, what they don't have in common. So the kind of basic methodological working assumption that all of Chinese historical phonology builds on is the Shesong hypothesis, which was articulated by this guy, Duan Yuzai, who's a kind of, you know, big hero in the history of Chinese historical phonology. And he proposed that characters with the same phonetic determiner are used to write words that would have rhymed in the language of the Shenzhen. And as I said, this is the cornerstone of all progress in Chinese historical phonology. I will talk about it a little bit more so that you have a good sense of kind of what it means, you know, for Duan Yuzai and for us. Which is that let's just think about the pieces of the claim. Characters with the same phonetic component write words that would have rhymed in the Shenzhen. So if we look at a series like this, it means that I'm trying to think of like what's evidence and counter evidence for this hypothesis. If the first and the second character in this series rhymed in a particular poem in the O's, that would be positive evidence of Duan Yuzai's idea. But of course, if two characters in this series occur in different poems in the Shenzhen or don't occur in the Shenzhen at all, then it's not relevant, right? So that's why I, it's phrased as would have rhymed in the language of the Shenzhen rather than rhyme in the Shenzhen because you'll have all kinds of, you know, just sort of holes in the data if you like, yeah? And similarly, most of the rhymes, many rhymes in the Shenzhen won't involve characters from the same Shizhen series because for one thing, they'll, you know, all of these words are some kind of ba or bie, yeah? Pa, pie. So you would expect that they would rhyme with something like ka, kie, yeah? But those would be from a different Shizhen series. So that's the, yeah. So I sketched that a little bit in terms of also I think why it has to, in a sense, remain a hypothesis, a sort of methodological research program is that there definitely are a lot of examples in the surging of words from the same Shizhen series that rhyme, which is why Duan Yusai came up with this idea in the first place. But, you know, there's no way to prove it just in terms of the available evidence. There's, you know, one way to think about it is if we had an infinite supply of old Chinese poetry, then at the limit, all words that are written with characters in the same Shizhen series would come up somewhere in rhyme position in such a way that they were connected with each other, that they rhymed, yeah? But of course we don't have an infinite supply of Chinese poems. So we have to live with what we have. Those are tones. And this is part of Baxter's system for writing Middle Chinese. So Chinese has four tones. If there's no capital letter and no consonant, it's the level tone. So let's say the first character is in the level tone. And the second character with the final X, that's the... Sorry, excuse me. That's the shang shang, so the rising tone. And it's believed to come from probably a final glow stop. And that's why he sort of chose this way of writing it, is it sort of, you know, what it means is in Middle Chinese, this is the shang shang, but it sort of is a convenient representation for what we think it probably comes from Middle Chinese. And then similarly, the final H refers to the Chu Xiang, which is the entering tone. Oh, sorry, not the entering tone. It's the... Someone out here falling, yeah, yeah. And then the entering tone is the fourth tone, which let's say from a kind of default perspective outside of East Asia, you wouldn't understand as a tone at all. You would just call it a closed syllable. Things that end in K, P, and T are called entering tone. Well, yeah, but I mean, you're getting... We're getting a little bit ahead of ourselves, aren't we, right? Like the... So Duan Yuzai had a hypothesis that is extremely influential methodologically, which is that two things in the initiation series would have been able to rhyme in all Chinese. Now, one issue that we will come across, the sort of the last question was also about is like, what does it mean to rhyme? Well, it means it tells you something about phonetic similarity, but also is a cultural thing, right? Like there's a famous example that I won't be able to bring to mind, but where Goethe systematically rhymes things that don't rhyme, let's say they are not phonetically identical in terms of vowel in modern German, but because Goethe... Probably I think people say in his dialect, there was a merger of those two vowels, but because Goethe did it, it's just considered like anyone can do it still. You can rhyme those two things, even though they don't rhyme. So I think we always have to be aware of, in rhyme as a poetic practice, that there's one element of it that has to do with, let's say real phonetic similarity, and there's one that has to do with cultural convention. So the question is, you know, maybe the, let's say maybe this last word is not part of this series, but part of like, let's say, another sub-series that's built on one of the characters here. So I think this is an extremely important point that unfortunately we're going to basically ignore, and the discipline largely ignores, which is to say there are several books, the most famous one is Carlgren's Grammatica Serica Vecenza, where he says, okay, we're going to organize all Chinese characters into about 1200 of these series. And subsequent developments in particular, Axel Schussler's book, which sort of updates Carlgren's, says like, no, some of these should be combined, or some of them should be split, but has the same approach, which is to say, there's only one layer of analysis. Two characters are either in the same series, or they're not in the same series. In reality, it's more complicated than that, that there's, it's like a network where one character is at the sort of center of a family of things built from it, but each of those can be at the center of their own families. This is, I think, one of the areas where, where a lot of progress in Chinese historical phonology is still possible. And Matias List and I have explored it a little bit in a paper that came out, I think, a year ago, but it's just a kind of proof of concept to sort of look at how you can represent these things and test different hypotheses, but it's really an area that needs more work. So thank you for drawing it to our attention. Okay, now we're going to move on to methods. So, yes, okay. So this is kind of a guide to, if you want to reconstruct all Chinese, you know, in the manner that it has been reconstructed so far, how do you do it? So you get your hands on some middle Chinese, like using rhyme books and rhyme tables, and then you internally reconstruct that language. So, you know, you just look for asymmetrical patterns and kind of hypothesize how, how might this language have been in the past. And you kind of reward yourself for those internal reconstructions that improve the rhyming in the shirching, and you kind of punish yourself for those internal reconstructions that do not improve the rhyming in the shirching. So then you end up with this kind of internally reconstructed middle Chinese that improves the rhyming in the shirching. And then you bring in the shesheng series by bringing your analysis of the shesheng series with respect to the shirching into conformity with the shesheng hypothesis. So putting it another way, like you, or just, yeah, putting it another way. So you develop ideas from within middle Chinese in such a way that they improve the rhyming in the shirching. And then you kind of test and refine and expand those ideas by seeing what they do to shesheng series. So that's how we reconstruct all Chinese, basically. So now I'm going to go through those. So you get yourself some middle Chinese. So we come up with initials and divisions in the rhyme books. Now I'm going to go through this, and this will rely on what we talked about yesterday. So if you want to know what the initials are in middle Chinese, you find these chains of Fangqi initial spellers. I mean, if you end up lost, let me know. But we talked about before when you spell a word, my example was bone equals bake plus phone. Now we know that bone and bake start with the same initial. So by looking up the Fangqi initial spellers in the shesheng, you can build up these chains. And this is two examples of doing this. So as you see on the left, we have a character chick. And then its speller is this next character. Wait, am I doing this right? Well, I'm not totally sure. But in any case, the point is simple, which is that the thing on the left is one chain you get. So if you look up one of these characters in the Cheyoon, it uses one of the other characters as its initial speller. And then if you look up that one, it uses it as its initial speller and you get this chain. So you know all of the characters on the left have the same initial. And similarly, you know all of the characters on the right have the same initial. Now at this sort of level of abstraction, if you're just looking things up in the Cheyoon, you have no idea, you know, is that initial a P or is it a G? You have no idea. You just know the things on the left have the same initial and the things on the right have potentially a different initial than the things on the left, but they have the same initial as all the things on the right. So now is where we bring in the rhyme tables, we bring in the footnotes where I point out that, so we have the Asterix footnote. This is this character that's pronounced something like Zhang. So if we look it up in the Yunjing, we're told that it has a voiceless African initial. And if we look it up in the Cheyoon, we are told that it has the so-called Zhou initial. So now we have some phonetic information. We can say, okay, well, we know that this second character on the left started with a voiceless African. And we therefore know that all the characters on the left started with a voiceless African, because we know they all have the same initial. And then if we look at the characters on the right, you see the dagger footnote, and you see that this character is on a different page of the Yunjing. And it's also described as having a voiceless African. And it's also described as having the Zhou initial. So this is a curious situation, right? Which is that the Cheyoon distinguishes these two chains of initials, which is why I've written them as C1 and C2. But the rhyme tables don't distinguish them. They say they're both voiceless Africans. So now we have to ask ourselves like, well, is it just a coincidence? Because it could be that the Cheyoon splits these two chains. Or is there some kind of merger that's happened between 602 and the 12th century? Now I'll go through and I think at the risk of being tedious, I think it's worth doing exactly the same exercise with this chart. So now we're looking at the first level of abstraction, just other initial chains. Because from within the Cheyoon itself, we have no information about how any of this stuff was actually pronounced. We have incredible information about the interrelationship of how different carriages were pronounced. But in order to actually apply a model to it in terms of articulatory or acoustic phonology, we don't have an in from inside the Cheyoon. So we look at these fancy initial-speller chains. And we get all the ones on the left are in one chain. And all the ones on the right are in another chain. And then we look up the characters and sort of an aside here. This also shows you that there are many, many, many more characters in the Cheyoon than in the rhyme tables. The rhyme tables only give one character per syllable type, basically. So in general, in a fancy chain, you'll only find one of its characters in the rhyme tables. So we try to look up the characters on the left in the rhyme tables. And we only find the last one. And then you look at the star and it says that the Yunqing says this is an initial liquid resonant. And the Cheyoon says that it has the so-called lie initial. Now, if we look in the right, we try to look up those characters. We similarly find only one of them in the rhyme tables, which is the second to last one. And then we find that it's also described as a liquid resonant in the Yunqing and as the lie initial in the Cheyoon Lui. OK. So how far have we gotten so far? We've looked at four initial or four fanche initial speller chains. And so these two and these two. And what we figured out is that these two chains are distinct in the Cheyoon, but they're linked in the rhyme tables. And we figured out that these two are distinct in the Cheyoon, but they're linked in the rhyme tables. Now, oh, I don't really go into it. But what we can do next is kind of try to come up with some theory about, let's say, the phonological system. And in this case, there, let me see if we have a clear example. Yeah. OK. Look at the second example on the left and the third example on the right. And you see that they have the same rhyme category, which means that there are minimal pair in terms of phonological analysis. Right? So we can say from within the Cheyoon, they have distinct initials, but the same rhyme. So they're a minimal pair. So then as a discipline, we say, OK, in this case, we will speculate that the Cheyoon had two kinds of voiceless African initials that were later merged in the rhyme tables. But if we look at this case, I leave it to the audience, to you. But there is no rhyme on the left and on the right in common. So this means that although the Cheyoon distinguishes two kinds of Ls, we can regard those two kinds of Ls as aliphons and not as phonemes, because there's no minimal pair, right? So in this case, if you like, we agree with the rhyme tables and not with the rhyme books. That's not maybe the best way to formulate it. But I think you see what I mean, which is one way of putting it is we start from the categories that the primary source gives us. And what it gives us is these onset-speller chains. And then we ask ourselves, well, what is the phonetic interpretation of these onset-speller chains? And we can do that with recourse to the rhyme tables, find out things like, oh, these two are voiceless Africans and these two are liquid residents. And then we want to ask ourselves, OK, but how many initial phoneme contrasts are there in Middle Chinese? And for that, we can look at the patterns of elementary distribution with the rhymes. And we can say, well, in the case of these two phoné initial-speller chains that we know represent voiceless Africans, they contrast in Middle Chinese. So they're two different phonemes. And when it comes to these laterals, we know they don't contrast, so we don't need to call them two different phonemes. And if you apply this kind of analysis across the board, let's say at least one solution you might come up with is this one, which is the standard solution. So these are the initials of Middle Chinese. We have, Qing is clear, right? So it has a kind of, let's say, so each of these words has a non-technical meaning, right? Yeah. So where I've said voiceless, the actual term is clear, and then the one for voiced is muddy and so on, right? And the question is, how do we kind of know what those terms mean? I think what I would say is that the space of options is not vast, right? Like, how many manners of articulation are there in attested human languages? We somehow have to map those on to our analysis of what these terms meant in Chinese historical, in the philological tradition. And an analysis that is kind of, I don't know, less exotic would be preferred. So if someone said like, oh, I think that the muddy initials are implosive and the Qing initials are adjectives or something, I would say, well, you have to, you know, the burden of, you know, the onus is on you to argue for why you think those are implosives or those are adjectives. Well, I would say this is like in this particular case, it's because from the other book, we have it saying lai mu, right? So that's where you kind of take just a little glance at Chinese dialectology and you say like, well, in almost all Chinese dialects, the character lai is pronounced with a lateral initial. So, and then you say, is there a reason why the yun zhing would call it clear muddy? And I think, I mean, so I think what I'm about to say is right, which is, I think it's a standard analysis, but basically the resonance didn't fall as, because there wasn't a contrast in voicing in Middle Chinese in resonance, they couldn't, they didn't, you know, split into clear and muddy, analytically speaking. So they said, ah, it's the clear muddy. And then used the fact that there wasn't a distinction in that class as a name for that class. That's I think the reasoning. But one thing you kind of have pointed out is that the interpretation of these categories does, you know, cast an eye at actually how characters are pronounced in Chinese dialects. And in like a Bernard Carlgren is the person who really worked on this in terms of like moving in the 1920s, moving from purely talking in categorical terms, right? If you're a qing phonologist, you don't care. You say, oh, this is the lai mu and you're done. Whereas if you want to think about the articulatory or acoustic phonetics of this stuff or write them in an alphabet, you have to, you know, tie into international analyses and Carlgren's the first one who really did that. And he did it with reference to Chinese dialects and particularly I think Sino-Japanese readings. Yeah. And I think in my presentation, I'm trying to emphasize how little you need to do that in order to show kind of the independence of the philological sources as sources. And I think I'm doing that in part to answer the kind of critique of that we talked about yesterday of Polly Blank and Koblin, which is like, well, look, if you have to look at dialects anyhow, you might as well actually just apply historical linguistics to the dialects in a rigorous comparative way. And I think that in order to sort of counter that critique, I'm trying to really emphasize how little you need to look at modern pronunciations of Chinese. But of course, we don't want to develop an analysis of the Chinese philological sources that are wildly inconsistent with the reality of how Chinese is spoken on the ground. How can I say it? I'm just going to present the orthodox view on this, which is that we know that Middle Chinese only had four tones because the Che Yun is organized into volumes by those four tones. And also, he's responding to an earlier research tradition of analysis of how many tones there are. And this is in my book somewhere, but I think that the categorization of Chinese as having four tones starts in around the second century. So what the field thinks is the reason why tones weren't mentioned before the second century is because Chinese did not have tones before the second century. And the reason why Chinese was discussed as having four tones from the second century on, but let's say for our purposes, from the second century into the Sui dynasty when the Che Yun was written, is because in fact Chinese had four tones during those centuries. So you're completely right that if you look at the phonetic reflexes of characters in these categories in modern Chinese dialects, in most cases the Qing Juo distinction will have something to do with tones. But what we think is that's a secondary effect having to do with tonal splits induced by differences of voicing, of manner, of manner contrasts on the initial. Because you will usually get, let's say if you look at the standard analysis of these things, you will usually see more than four tones and that it will be possible to analyze all those dialects that have a Qing Juo tonal contrast as having in some sense eight tones with possible neutralizations. So I think that's strongly kind of indicative that the orthodox tradition that I'm pointing at is right. There are people who analyze all Chinese as having tones. But I don't think there's anyone who analyzes all Chinese as having tonal contrasts along the Qing versus Juo distinction. There are people who analyze middle Chinese in that way. Okay, so now we want to come up with the divisions. Okay, so I've been doing something that I hope everyone follows, which is I translate Deng when we're talking about the rhyme tables as rank and when we're talking about the rhyme books as division in order to emphasize that they are not the same thing. And I think it's extremely confusing that the Chinese uses the same term for them. And this is not my idea, it's not my choice of terminology. It comes from this fellow, Oh, who wrote the I think wrote the article in the Brill Encyclopedia of Chinese Language and Linguistics about division. And then just a kind of word of warning that there's a new book out by Zhong Weicheng about Chinese historical phonology with Cambridge where he uses the terms in exactly the opposite way. So he uses division for the rhyme tables and he uses rank for the rhyme books. And I think what a catastrophe, like why did he do that? But let's say I'm sticking with the way Oh did it. And I in any event think that it's a very useful distinction. So let's now explore what that is. We've already discussed that in the in the in the rhyme tables division, you know, I'm taking as just means something physically about the page formatting of the book itself. Yeah. So now we move from rank to division. So step one. This guy, Zhang Yuan in in the in the in the 18th century, mostly. He looked up the characters in each Che Yun rhyme category to see which rank the rhyme tables put them in. Okay. And what did he find? He found that the rhyme categories of the Che Yun divide into four classes. Those rhymes that contain characters appearing only in rank one. Those rhymes that contain characters appearing only in rank two. Yeah, and then here's where I was going to say, now you think, oh, this is pretty going to be pretty uninteresting. Yeah. No, it's gets more exciting when we get to the third category. Those rhymes that contain mostly characters that appear in rank three, but also contain characters appearing in rank two and rank four. And those rhymes that contain only characters put in rank four. So these are the divisions of the Che Yun. Right. And you see how they are very explicitly different than the ranks of the yunjing. And why are they different because of the messy third category. Something that's division three could be rank two or something that's division three could be rank four. Okay. And yes, and then I want to make this very important sort of analytical distinction, which is that there are no divisions in the Che Yun. Right. There is a hybrid sort of analytical beast that combines rhyme categories of the Che Yun with the ranks of the rhyme tables. So, like, you think about this, even in terms of chronology, like they're this kind of entanglement of information from 602 with information from the 12th century in this very complicated way. So I think it's just necessary to keep that very squarely in mind because you are always tempted to sort of reify these things and say like, well, how was the second division pronounced. Like, well, you know, you can't have a thing that was pronounced in two different centuries at the same time. So, so we need to keep in mind how constructed these categories are as abstract analytical categories. Yeah. So that's step one of coming up with the divisions. Now, the realization that the divisions, as we've just defined, mostly coincide with distributional classes of initials allows one to redefine the divisions using co occurrence patterns of initials, rather than referring to the rhyme table ranks, and say, but from a methodological perspective, as I just explained, I am unhappy with this definition of the divisions, because it's this weird combination of, you know, information about the page formatting of books in the 12th century, and kind of the distributional or and rhyme categories in the Che Yun. So I'm trying to find a way of paraphrasing the categories found in entirely Che Yun internal manners. Yeah, it turns out we can basically do that by making reference to initials. So let's start from from the from the Che Yun perspective by making reference to a fancy initial speller chains. Right. So let's do it. Yeah, the thing I want to emphasize first because there's actually a kind of mistake in my book that that the zev handle points out in his in his review. I'm going to we're going to be looking at like what look like complimentary distributions, but they're only complimentary distributions because we're restricting the class of initials we're looking at so we're only looking at the dentals and the frecatives, sorry, the the retro flexes, we're not looking at the dealers and the labels. Yeah, so, so I that's an important caveat for me to put in that I didn't have in the book enough and made a mistake, but let's just say we're looking at the compatibility of the divisions as we've just defined them, in this way I'm unhappy with, with current currents patterns of different initials. Okay. And then you see what you see, which is to say like the Laimu occurs in division one, three and four, and the dentals only in division one and four, and the retroflex in division two and three, and the africate in division one and four, and the palatals only in division three, and the palatal retroflexes only in division two. And the point that I want to make right now is, okay, this may not yet seem super elegant or meaningful, it's just the fact about the world, yeah, but it's already allowed us to define the divisions with respect to the Cheyum, right? Because rather than defining division four as those characters that occur in rank four in the rhyme tables, we can define division four as those rhymes that co-occur with the dental and africate initials, but not the retroflex initials. You see that? Like, you might well ask why we want to bother doing this at all, yeah, in terms of like, why do we want these divisions, yeah? But let's set that question aside for a minute and instead say my goal kind of methodologically speaking is to try and pin our theories of all Chinese historical phonology to the oldest, most comprehensive sources. So I want to kind of leave the rhyme tables behind and only stick with the rhyme books and these distributional classes allow us to offer new definitions of divisions in Middle Chinese that do that with an exception, however, which is, you notice that the dentals and the africates cannot be, or the divisions, yeah, yeah, here's how to put it. The division one and division four cannot be distinguished on the grounds of co-occurrence classes with different initials. So that means that, let's say, we can forget about division, the distinction between division one and four and say that there's kind of only three divisions, yeah? And that they are division one slash four, division two and division three. But if you just do that, and I sort of ask you to do it, I should have had an extra slide, but I've asked you to do it in your mind, you say, okay, then I just, let's say cross out the first row. What you then notice is that division two and division four are in complementary distribution. So we can, in terms of co-occurrence classes with initials at no loss of information, replace the analysis of four divisions with only two classes, type A and type B. So I'm just gonna paraphrase that, which is from the perspective of the distributional classes of the Cheyoon itself, which is Middle Chinese, the analysis of four divisions and the analysis of two types are both compatible. Now, and it's basically a matter of taste, which one you want to go with. But as a fact about the way the discipline works, basically everything after the Cheyoon, we talk about divisions, whereas there's a tendency when talking about things before to talk about the A, B distinction. And that makes sense, right? Because the source material for Middle Chinese kind of presents, especially the rhyme tables, presents these four ranks. So people talk about, oh, well, how is rank four different than rank one? But the further back in time you go, the divisions and the ranks don't matter because all of the contrasts that we're talking about can be understood in a simpler terms as the A, B distinction. So for the rest of the course, we'll be hearing a lot about A versus B, but not that much about the ranks and divisions. Okay, yes. I think we can do that, but then we might not be able to have the A, B distinction. Yes, I mean, actually what I was thinking about, which is the kind of thing you're talking about, but not the exact example you're talking about, is we do have examples of Laimu with division two. I think two characters, I think or something like that. So the, just a second. So what I would say is, yes, we ignore stuff like that. Like if something's sufficiently rare, we ignore it. But you don't ignore these things all the time. So for instance, I just happened to know about, because Gongshun has written a blog post about it. One of these examples of a division two L initial character is actually the Lan in An Lushan. Sorry, the Lu in An Lushan. So An Lushan was probably a Sogdian. And you get, so which is to say, An Lushan is not a Chinese name. It's probably actually the same name as Rock Sam, which is a Sogdian name, yeah. So these, so there is also a tendency that these kind of really weird characters, like characters that occupy basically empty phonetic positions are late in terms of the epigraphical record and are used to write foreign words. So I mean, let's say, don't take my word for it. That would be an interesting thing to read a paper about or something like, what is the philological dimension of kind of characters that represent what should be non-existing syllables. But I would say my impression is that overall, it's okay to ignore them, not just because they're rare, but because they're rare and late and tend to be used for exactly those situations where someone might find themselves trying to write something impossible, right? Like for instance, in English, we would say, Q always is followed by U, except when it's not, and when it's not, it's being used for Arab words, for example, because we're trying to use the Roman alphabet to represent sounds that aren't normally represented. That's a sort of socio-historical question, but what I would say is you, let's think of it this way. Could you motivate the type A, type B distinction without ever mentioning the existence of rhyme tables? That would be fun. That would be a fun thing to try to do. And if so, then I'm happy to agree with you and say that the discipline of Chinese historical phonology with respect at least to the very origins of the language in Sino-Tibetan to 602, we should only talk about divisions and forget, sorry, only talk about types and just totally forget about divisions. I would be okay with that. I mean, some of you are immediately thinking like, but you're gonna talk about division two and the R hypothesis and whatnot. Yes, we'll get to all that, but I think like, well, maybe it can be done without analysis of divisions and maybe that would make for an overall more elegant presentation. I mean, I think there's a tension in all academic disciplines, but I would say especially in all Chinese historical phonology of how much do you stick with tradition? Because on the one hand, if you don't stick with tradition at all, your analysis in your writing will be unintelligible to other people working in the discipline. But on the other hand, I think there's a tendency, or I mean, so I apologize for a slight digression, but I was a math major for a while. And one thing that I noticed in math is like, the moment someone comes up with a better, shorter proof, everyone ignores all previous work on that problem. Like the discipline loves to forget about its own past, at least in terms of how it's talked to undergraduates. And that made me feel very sad as a mathematics undergraduate because I think of all these poor mathematicians who pour their lives into slightly less elegant proofs who now we ignore, right? But when I started doing Chinese historical phonology, I felt like we have the opposite problem, which is we make all students re-experience every mistaken idea and every inefficient formulation since the middle of the 19th century until now. So there's all this discussion about like, well, Carlgren was wrong to distinguish four types of whys or something like that. It's like, well, who cares? Let's talk about what's true rather than what Carlgren was wrong about. And that is one of the things I've tried to do in my book is to kind of by focusing in on the primary sources to kind of forget some of the baggiest of the history of the discipline. But as you can tell, I've only been semi-successful at that and look forward to us all together collaborating on finding the most efficient, concise, clear and empirical presentations of the actual knowledge of Chinese historical phonology that we have. Let's just say I entirely sympathize. Yeah, this is just kind of what it's like doing Chinese historical phonology. But I would say I'm trying to do two things. I'm trying to kind of make it clear what structures are imminent to the primary sources. So we can say like, in the rhyme tables, there's a thing called ranks. And we can say like in the rhyme books, there's a thing that's actually part of the presentation of the document homophone groups. And then to explore the less directly observable but still genuinely their categories. Like what are the relationships of complementary distribution between different classes of initials and different classes of rhymes. Those are facts about the presentation of the language that were being presented by the Cheyoon that the Cheyoon itself didn't showcase. So I would say one point is just to figure out what structures we have there. And that we can understand as a kind of, I don't know, scientific project in its own right, setting entirely aside the question of the phonetic interpretation of these categories. That's a good way of putting it, yeah. We're looking at the phonotactics of middle Chinese while trying to be quite open-minded about the phonetic interpretation of the categories we're working with. Now, I think that there's a more abstract way of doing this than I've even done it because I've said sort of capital T, whereas that's not in the book. But I think that's safe in terms of like, it's not sort of overly artificially realistic and can just be thought of as an arbitrary label for that category. But it's a one we've chosen for heuristic reasons, yeah. Now, the question you might well be asking is like, why don't we come up with terms for the divisions and the AB distinction that also have that kind of heuristic of implying of phonetic interpretation? And I would say that, well, you're welcome to do that. But what we are coming up to slowly is what are the phonetic interpretations of these categories? And before we determine that, I think it's good to have the indeterminacy in focus. And like, let's say, this is a, I don't know, this is maybe kind of we're just counting the angels that are on the heads of pins here, but like Yeom Zhak likes to do everything in the IPA. He writes middle Chinese in the IPA. He talks about the interpretation of the initials in the IPA. And I think that that is like, is fine, you know, it's not gonna, no one's gonna die from it, yeah. But it has a tendency to seduce us into thinking that things are more knowable than they in fact are. And it tends to bring out of focus the labor necessary on all of our parts to know that we know what we're talking about, if you like. With all of that said, there's also a concrete issue which is the AB distinction actually, you will be unhappy to hear. We don't really have a good handle on what its phonetic interpretation is in all Chinese, but it matters immensely. Like all sound laws basically operate differently if you're talking about type A syllables or type B syllables. So you just can't formulate any sound changes in the history of Chinese without reference to type A and type B. When I just wanted to indicate this kind of internal reconstruction of middle Chinese that I was talking about where we can notice things like, there's no, I'll call it H although it was probably actually a gamma in type B syllables and there's no G in type A syllables. So then we can just speculate that the H is some kind of united form of the G and I'm presenting this at this moment just to show you kind of the kind of internal reconstruction we do not to propose this concretely, which I'll do when we go through the initials of all Chinese and just give you another indication of doing the same sort of thing in terms of improving rhyming in the shurjing. So this is a hypothesis that a certain vowel in middle Chinese which you can think of as a comes from a ra. And then I will just point out that in Ode 2.2 we get a rhyme between aak and aak and in Ode 3.3 we get a rhyme between aang and aang. So the hypothesis that a comes from ra would improve this, yeah? So that's again just going back to the very beginning just internally reconstruct middle Chinese in a way that improves the analysis of the shurjing. And now bringing the shisheng series into conformity with the shisheng hypothesis. Oh, I don't give an example, oops. Okay, and then this will become very methodological but actually some of the questions sort of some of the questions are touched on which is do we really have to do things this way? Like it seems a little bit tortured. And I just wanna point out that in mathematics we oftentimes define the attributes of something. For instance, the integers one and two and three they're a set that's closed under addition and subtraction, for example. And then we can abstract away and look at all objects that have those properties. So I want to suggest that as a discipline we take this approach with historical Chinese phonology. So now we look at what, oh, hurry over time. Just stick with me like a few more seconds. What are the roles that the evidence play? So the rhyme, yeah. This actually isn't quite right but the rhyme tables, what is their function? Their function is supplying phonetic interpretations. So I call that alphabetizing. But they only do it for very, very few characters. So what is the purpose of the rhyme book? I call it transitivizing. It's telling you which other characters you can apply those same interpretations to. Okay. So then I just want to point out that there are other primary sources we could look to for these two actions. For instance, Han Dynasty Chinese transcriptions of Indic Buddhist terms could be used as alphabetizers or Tang Dynasty Tibetan transcription, Tibetan alphabet representations of Chinese words, Pakpa script in the Yuan Dynasty. These are all things that can be used as alphabetizers. And then there are other transitivizers like Peron, Peronomastic glosses from the Sherming and Fanche from other books from the European, from the Jingjian Sherwen. And I think that all Chinese phonology has too much obsessed on specific alphabetizers and specific transitivizers. And we should systematically explore what happens if we use different ones and make things more explicit like who's analysis of Shijing Rhymes and who's analysis of Shesheng series. And now of course this becomes overwhelming in terms of data, which is why we should use explicit formal models and digital representations of these sources. And so that's kind of where I wanted to end in terms of sketching a kind of future of Chinese historical phonology where we explored in the same way that has been done for the rhyme books and the rhyme tables, the complete body of available evidence. I think that's a methodological thing, which is like if you have two assumptions, go with the simpler one. So find me and a circumstance in which it would be beneficial to understand the initial fangche-speller as containing a minor syllable and I'll be convinced. I think that particularly the earliest fangche would be a place to look for that, that'd be fine. But I do think that like it's clear that by middle Chinese we're dealing with monosyllabic, let's say monosyllabic morphemes, although the morphological analysis isn't the point. The categories that Che Yun works in are monosyllables. Let's put it that way. I think that's pretty clear. The short answer is no, and I'll just tell you my own feeling is that we shouldn't do that. And very succinctly, I would say, at the moment computers can do two things. They can logically reason by symbolic manipulation. And they can have intuition and that's what artificial intelligence is, is intuition. You show a computer, you know, millions of pictures of penguins and then you show it a new picture and you say, is this a penguin? You have no idea how it's making a decision. You just rely on its intuition. And the way that human beings are still different than computers is a human being can switch seamlessly between logic and intuition and can kind of try to pin down their intuition and logic and whatnot. Whereas a computer, you can either ask it be logical or you can ask it be intuitive. And at the moment, I think Chinese historical phonology would benefit from more logic and less intuition. So that's why I think things like finite state transducers are useful is to actually make our hypotheses extremely explicit and test them symbolically to force us to do things like say exactly which order sound changes happened in. And I think for that sort of thing, artificial intelligence is just not helpful. The answer is yes. So the point that, gosh, I'll try to put it, the point that I'm trying to make though is that that distinction, if it's in the Cheyun, cannot be recovered kind of kwa division. Does that make sense? Like that if you're trying to paraphrase ranks as divisions, you cannot separate the first and the fourth division using Cheyun internal evidence. So it's just a fact about what information is in the Cheyun and what information isn't in the Cheyun.