 Today I will talk about the phonetic information contained in the Chinese script. So just to start, you know, very simple, right? We all know about these different, these four words for ma that we use to teach tones. So mother is ma and horse is ma and mother as a character is made of the two components so on the left is the part that means woman and on the right is the part that looks like it means horse but here it's saying pronounced like ma, right? So this is very simple. And then in a way we can say, you know what's the information encoded in the character? The information is say the morpheme that sounds like horse and has something to do with a woman, right? And then, you know, in my own sort of thinking about this you have to know it, right? It's the structure is a kind of aid to memory but you have to know, oh, the, because I know Chinese already the word that has something to do with a woman and it's pronounced like horse is the word for mother, right? So just to say from now on, so I'm going to use middle Chinese for the pronunciation of Chinese characters, not the pinion because it's a little bit, I don't know because I'm more familiar with it frankly but also it's further back in time. So middle Chinese is from 602. So it, and then the characters of course were invented you know, I don't know in the Zhang or the Zhou dynasty. So middle Chinese is the earliest point where we have totally systematic information about the pronunciation of Chinese characters. So I'm putting pinion behind us and we go with middle Chinese from now on, yeah? So now we look at the series of characters written with the phonetic horse and the question we're exploring is, what information is it that we're being given by the fact that a particular character is written with this phonetic, yeah? So we have this character, yeah? And then this one and then this one, oops just I think it's just the three of them in this series. So it's quite a small series. And now we ask ourselves, so what does this phonetic mean? And we can say for sure that it doesn't tell us anything about tone. The first character is shang shang and the second character is Zhu Xiang, I think that's right. And then the third character has both readings, right? So clearly the information, what tone is it is not indicated by the phonetic. But from this example, it looks like maybe the phonetic is telling us what the initial is and is telling us what the vowel is, yeah? Or the rhyme, if you like, yeah? So ma, so now this is our initial hypothesis that the phonetic, I mean, this is a conceit, right? For the purpose of exposition, right? But our conceit is maybe what a phonetic tells us about the pronunciation of a word is its initial and its nucleus, but not its rhyme, yeah? So now we look at another series, yeah? This one based around bieh, so we have bieh, we have bieh in a different tone. We have bieh, we have bieh, we have pa. And then we have pa also in different tones and then pa with the aspirate and then ba, okay? So now we look at this and we think, okay, well, our idea that the pronunciation tells us nothing about the tone is confirmed. And our idea that the phonetic tells us the initial is sort of disconfirmed, right? Because here we have, I'll just go through it, we have b and p and ph. So clearly the phonetic is not telling us which one it is, but these all are obstinance with the same place of articulation. So maybe the phonetic in the character is saying bilabial stop, yeah? This, whatever character I write with this phonetic starts with a bilabial stop. Maybe that's what it says. And then as for the idea that the phonetic specifies the rhyme or the main vowel, we see here, we have two rhymes. We have a and we have yeh, yeah? So it seems that this was also disproved if you like this conjecture. But actually it's not the case because if we look at the odes, here is od 18 stands a one on the lamb furs, five many trests, sorry, many thread trests of white silk. We see that this bieh, which is in the Shixiong series we're talking about, so it has the phonetic we're talking about, rhymes with da. So that's because I'm writing this in middle Chinese, yeah? The bieh is in middle Chinese and the da is middle Chinese, but the poems are written in old Chinese. So it looks like actually maybe in old Chinese these were the same vowel, yeah? And just to give you another example from the same Shixiong series, this is od 145 stands one. By the shore of that march, there are sedges and lotus plants, where here this bieh it rhymes with ha. So again, we have another character in the same series written with the same phonetic that rhymes with a character that is pronounced in ah. So it looks like yeh and ah, rhymed in the ods. And then this means that here where it says two rhymes, ah, and yeh, these are actually somehow one rhyme, yeah? Okay. So it was based on evidence of this type that Duan Yuzai, who was a great Qing linguist, philologist in the 18th and 19th century, proposed what we call the Shixiong hypothesis. And that is that any two characters written with the same phonetic component, write words that would have rhymed in the language of the shi Jinping. Now it's necessary to praise this way would have rhyme because of course many characters, most characters in fact, are never used as rhyme words in the shi Jinping. So it's not a sort of testable falsifiable hypothesis. It's more like a research program. And the way to think of it is like, we only have 350 poems in the shi Jinping, but if we had an infinite supply, yeh? Then, you know, that's the sort of thought experiment. If we had an infinite supply, eventually we would find the character we're interested in and we would be able to verify that it does, in fact, like the two characters with the same phonetic rhyme in the shi Jinping. But because the shi Jinping is so small, we actually only have sort of indicative evidence of the type I've already shown you. And instead it becomes a kind of matter of faith. We just, we believe in the Shixiong hypothesis. This Shixiong hypothesis in the form of, that Duan Yucai has said it, you know, it is something that is really one of the basic tools of doing research in Chinese historical phonology. The other component that we recognize, this fact that this series has all of the initials are, to use the technical term homo organic, it means that they are articulated, the initials are articulated in the same place of articulation, but with different manners of articulation. That's what it means to be homo organic. So they're all bilabial stops. That observation was kind of brought into the Shixiong hypothesis by, explicitly by Li Fengkui, saying that it's just changing the observation into a kind of a doctrine if you like, right? So any two characters that share the same phonetic component should have homo organic initials. That's the, these are the two components of the so-called Shixiong hypothesis. So if two characters have the same phonetic, that means they have the same rhyme and that means they have homo organic initials. So in a sense, it means that each series, each phonetic should correspond with one syllable type in old Chinese. So they can have different manners of articulation and they can have different tones, but they should have the same rhyme and the same place of articulation for the initial. So now that we have this, these two components of the Shixiong hypothesis as like parts of our research program, we want to look at places where they appear to be violated and see whether we can fix them. Shixiong series that makes middle Chinese pronunciations with non-homo organic initials provide an opportunity to explain the divergent middle Chinese pronunciations as phonetically conditioned developments of all Chinese readings that were homo organic. Now we've changed these observations about what it seems like the phonetics are encoding into a kind of doctrine of what we presume they're encoding and then we use that to actually do reconstruction. So I'll talk you through some examples, but first I just have to introduce something that I'm not going to actually talk about which is called the AB distinction. And actually we've already sort of met the AB distinction in this series here, yeah, where I'll just tell you the characters with the reading in middle Chinese are, those are so-called type A syllables and the characters with the reading yeah are so-called type B syllables. So as we've already proven in this case, they, these had the same rhyme in old Chinese and then it split in middle Chinese. And so the conditioning factor for the split is the so-called AB distinction. I'm not going to talk at all about what this might have been phonetically it's extremely controversial, but it's clear even from this evidence that I've shown here that there was some conditioning environment for this split in old Chinese. And we call that the two settings of that conditioning environment type A and type B. And it comes from the, actually from the, let's say our awareness of it as a distinction and our way of talking about it comes from the Song Dynasty rhyme tables where, and this is just for those people who have some background in this sort of thing where the type B syllables are the third division, the san dung and all of the other divisions are type A. But if you're not, if you don't know about divisions in the Song Dynasty rhyme tables then you can ignore what I just said because it doesn't matter for this presentation. And instead, this slide is just there to point out that the AB distinction is itself written in to the structure of Chinese characters for at least some of the longer phonetic series. So in this series, which is actually in Carl Bern's numbering of phonetic series, it's series one, it's not in other people's but the kind of mother character of all of them is on the left and then it has certain characters that use it directly as a phonetic. But then on the right, there's one of its daughter characters that itself is used as a phonetic for a bunch of other characters. And then these are, let's say, the characters which are nodes in this graph to use kind of the technical terminology have been colored according to whether they're middle Chinese initials, sorry, yeah, you know, it could be, let's say rhymes is better, but let's say whether they're middle Chinese rhymes are type A or type B and the ones on the left you see are type A syllables and the ones on the right are type B syllables. And you can see that these are written using Baxter's transcription middle Chinese which are a little bit different than mine but you see all these J's, yeah, those are yuh which I'm writing with an I and that means there's something palatable going on in those syllables. So all of the syllables on the right are type B because they have this J in them which in my system would be an I. And you see all the readings on the left, there's no I, there's no J, yeah. So just to reiterate to sum up, at least with this phonetic series and there's a few dozen such phonetic series, the phonetic series itself breaks into kind of two components, the main series and the sub series if you like and the sub series is specifying type B which means that whatever this A, B distinction was the people who first invented these characters were aware of it and were using it in their decisions about making Chinese characters. So now I've introduced the type A, B distinction and proven that it's existed for a long time. That's the whole point because it's going to come up for the rest of my presentation. So now we're looking at just a phonetic series where the initials are not homo-organic. So Cha is a palatal in the first one and then Ha is a retroflex in the second one and Da is a retroflex and then we have two dentals. So this series using these middle Chinese readings violates the Sheshen hypothesis. So now we want to fix it somehow, right? So the first proposal we'll throw out there is that all Chinese T change to Cha in type B syllables and then the second proposal that we'll make is that all Chinese Tra change to Cha. And this happened both in type B and in type A syllables so we don't need to specify. And similarly that Dr changed into Da. So those are the proposals we're gonna make and then also mentioned that these same proposals can be motivated on grounds that have nothing to do with phonetics, like phonetic components of characters but entirely based on distributional patterns inside middle Chinese using internal reconstruction. But that's just a sort of side note to say we have independent motivation for these three proposals. So now if we, did I write them? No, I didn't. But anyhow, now you can imagine the first one started with a T, the second one with a T, the third one with a D. So now they're all homo-organic in old Chinese if we believe these proposals, right? Okay. So now we look at this series and here we have other Ds. So yeah, so the third one is D, the second one is D. So no problem, right? Because we've already found a way of changing retroflex initials into dentals but what about this Ya at the beginning? It's a problem, yeah? So this is what I just said that the Da can be taken back to Dr but that only gets us so far in making this series conform to the Sheshan hypothesis. So the question is what sound do we know of, in linguistics in general, maybe, that can become both a Ya and a Da? And the answer is L, yeah? And what I should do now is give you lots of examples but I haven't sort of done the homework to do that but I will just point out that in Spanish, Ya, like in mayyamo, comes from L, yeah, it's still spelled L and then in Latin it actually comes from a cluster. So in this particular word, clamare is what is where yamar comes from. And then L can be fortified into D. That happened in Latin as well but I don't know the details and it definitely happens in Tibetan, yeah? So we have good reason, just if we pose the question, what could become a Ya sometimes and a Da in other circumstances and L makes sense? So let's go with that and this was originally proposed by Edwin Pulleyblank. I'm slightly simplifying. He didn't quite propose it in the modern form, yeah, we're in the form we currently use and then Baxter and Sagar have taken over that proposal from him in exactly the way I'm formulating here. So L in type B syllables becomes Ya and L in type A syllables becomes Da and L followed by an R, whether in type A or in type B becomes Da. Now I'll just mention that this is Baxter and Sagar's notation and I probably should have not included it because it's probably confusing but this little pharyngealization symbol means type A. So they're sort of very tentative proposal is it maybe type A syllables were pharyngealized and type B syllables weren't pharyngealized but you can just see it as a kind of almost arbitrary index that says that if that little pharyngealization symbol is there that means it's type A, yeah, okay. So here we've done it, we fixed this series, right? So it all three started with L, yeah. So now moving on here, we have the problem that the series has one reading with a velar nasal initial, and one with a ha, it's actually, it's like a Chinese ha, it's a voiceless velar for good. So how do we change, I mean, this is what we're gonna do, you could maybe say the reverse but how are we going to change this ha by going backwards in time into some kind of nasal, yeah. And then similarly, we have series like this one and this one's clear because the majority reading is nasal, we have velar nasal, nga and nge, and then we have this other character whose reading is he, so probably the he comes from some kind of velar nasal, right? So this is the proposal that the ha readings come from voiceless nasals. So all Chinese had something like nga and the nga became ha, whereas the nga stayed nga, yeah. And that way it's just like the variation we see between P and B, right? It's like we have voiceless labials and voiceless labials in the same series. Here we have voiceless velar nasals and voiceless velar nasals in the same series. And then in this series, similarly, we have nga, we have nga and then we have ha. So I've given you one type A example and one type B example of voiceless velar nasal changing into ha. Okay, now we'll just go through and I'll maybe speed up a little bit so you don't get bored. The other places of articulation with the nasals. So in this series, the first one at the top, we have two readings, haq and m, or haq and mak. So the proposal is that the haq comes from mak, yeah. And then that's in type A syllables and then in type B syllables here, we have two readings or two characters, one with the reading schwet and one with the reading yet. So we think that the first one, probably the H came from a voiceless m, ma, yeah. All right. And then same thing for the dental nasals. So we have this series with the readings tan, nan and nen. So where does this tan come from? So so far, the poorly behaved readings in nasal series started with an H. Now I have a poorly behaved reading that starts with a th. But in case ta and, well, ta and na actually have the same place of articulation but we like nasals to have their own series. Let's put it that way. So in this case, we think that tan came from sna, nan, yeah. And then looking at type B syllables, so we have these three characters that in middle Chinese are nia, nia and shia. So now the odd man out is the shia reading. So why not use the same proposal? We say shia comes from nia. And now you also see, let's say methodologically, the nice, the convenient thing about the A versus B distinction is I can propose two different sound changes, right? I can propose it, nia changes into ta and I can also propose it, nia changes into shia because one happens only in type A syllables and one only happens in type B syllables. So there's no contradiction to the exceptionalistness of the proposals because of the A, B distinction is a conditioning environment. And then last but not least, this ha, we can reconstruct back to nia and that occurs in type B syllables. Whereas you know, we might have expected it to be a shia based on the middle examples. All right, so now we've dealt with, so just to review, we started with dentals and just showed that we can, this is the kind of simplest case, we can fix a series like this. So first we fixed this kind of relatively simple series. Then we fixed this more complicated series and thereby proposed a new kind of consonant in all chains, laterals. Then we proposed voiceless nasals in order to deal with these examples. And now we're going to propose other voiceless resonance because kind of once we've let the cat out of the bag and we're letting ourselves have voiceless resonance, maybe we can have other types of voiceless resonance. So here is some evidence for voiceless laterals. So we've already shown that a series that mixes d and l, sorry, d and y, yeah, we will reconstruct back to an l, yeah? But now we see two other readings in this series. So if you only had the first two, you would say, okay, fine, this is l, this is very straightforward l series. But then we have this shoo reading and this two reading, what to do about those? So once again, we just say, well, how about a voiceless lateral? A voiceless lateral that in type B syllables becomes shuh and in type A syllables becomes ta, which is the same behavior that we got with the voiceless nasal, the voiceless dental nasal, right? So that seems like a reasonable proposal. And then we can also find cases where we want to do this with r. Now, I actually haven't discussed, but middle Chinese l comes from r, we have reasons to do that. And partly, we want the l for other things, right? For explaining the mixed d, y series, okay? So a series where you have le, le and te, it looks like we're doing a sort of le thing here, but then this te gets in the way. But then we propose that shuh in type A syllables becomes ta and then in type B syllables, we propose that shuh becomes te, te, yeah. So that's interesting. This sort of breaks the symmetry a little bit. But anyhow, there you go. So here are our voiceless resonance and then just a little bit of intellectual history. The first voiceless resident that was proposed was the voiceless nasal, which was proposed by Dong Tong He. And then, you know, some people liked his idea. So then Polly Blank extended it to the velar nasals, the labial velar nasals and the dental nasals, as well as the voiceless lateral. And then Baxter added the voiceless rhodic. So yeah, I sort of mostly presented them in the order that they were proposed, although I started with, I think with the velar nasal and maybe I should have started with the labial nasal, yeah. And then, so Baxter proposed the last one in 1992. And then my reconstructions, I've been following the system of Baxter and Sagar from 2014, so they've accepted these proposals. Okay, and then here's just a summary chart of the voiceless resonance. So you have nga, ma, changing into ha, and then you have na, la, and shah, changing into ta, in type A syllables. And then you have nga, and ma, changing into ha, in type B, same as in type A, no difference in between them. And then in type B, you get nga, and la, changing into shah, but you get shah, changing into ta, ta, right. So I think, let's say, how long have I been going? About half an hour, yeah. Well, no, okay, well, I have lots of time. So then I'll also show you the uvulars, but I'll say that, you know, if you found, if this is started to get boring, then, you know, you can come back in five minutes or something like that, because it's kind of more of the same, yeah. But the problem is the same, the motivation is the same. Fixing poorly behaved phonetic series, yeah, okay. So some series mix velars with a glottal stop. So this first row is velars, right? K is a voiceless velar stop, G is a voiceless velar stop, this h with the line under it is a voiced velar fricative, and then the normal h is a voiceless velar fricative, yeah. And then we have this glottal stop, these are all middle Chinese initials, and then we have the yi initial also in middle Chinese, yeah. So yi is a palatal, the glottal stop is a glottal, so they're not velars, that's the point. So if we have a series like this, you see that it mixes velars and glottals. It has three velars, two fricatives, one stop, and then a glottal. So this is violating the shesheung hypothesis. So what to do? How can we deal with this? So how about we propose that the glottal comes from a voiceless u-vular stop and that the h, so the velar fricative comes from a voiceless aspirate u-vular stop. So I will try to make these, but I'm no phonetician. The q is like in atar, the country in the Middle East, yeah. And then the qh, you just put a little bit of air after that atar, atar, yeah. Yeah, so this is the proposal we're going to make and we see that, I don't know, this is what it looks like if we add that proposal. We haven't quite gotten there because we haven't dealt with the fourth one yet or the first one for that matter, yeah. But we can add this proposal, basically for purposes of symmetry, right? If the voiceless velar fricative comes from a u-vular, maybe the voiceless velar fricative also comes from a u-vular and that's a voiced u-vular, which is a glott, yeah. So I'm not going to get into Baxter and Cigar's explanation for the velars that pop up in u-vular series. It has to do with consonant clusters and I just wanted to keep this simple with the simple initials. But basically you can imagine that when u-vulars occur in certain kinds of clusters, they change into velars, yeah. And also typologically, I hear you would have to look at Pongmian or Arabic or something like that, but for u-vulars to change sometimes into glottals and sometimes into velars is perfectly normal, right? It's basically they're quite far back in the throat u-vulars and then if they move a little forward, they become velars and if they move a little further back they become glottals, yeah. And then we also have series that mix ya with velars like this one. And here, it's a sort of fourth problem if you like but we can use the same solution by proposing the voiced velar, sorry, the voiced u-vular stop G. But what's the difference? Well, here the G is in a type A syllable and here the G is in the type B syllable. So again, the A, B distinction is helping us to do work for us without adding a whole lot in terms of machinery, yeah. And then, this is just a little detail, but before front vowels, Baxter and Sagar think that GW, so a voiced labio-u-vular becomes ya and they point to this series as evidence. So you see there's an N, E, front vowel. And then otherwise it becomes a velar fricative in type B. So this is another way to get, they've noticed that you can get the two different outcomes with a condition change. So you don't have to propose five, six different sources, yeah. So I think skip over this one, yeah. And just say that now in terms of the intellectual history, this u-vular hypothesis was proposed first by Pan Wuyun, but then Baxter and Sagar have slightly altered his proposals and I won't go into those details, yeah. And then just to point out that there is some comparative evidence that suggests that this is on top of the U-vular narrative evidence that suggests that this is onto the right track, basically foreign words in early Chinese. These are military titles. So the first one is a Zhongnu military title and Alexander Woven thinks that it is borrowed from Proto-Yenesean where, you know, say for totally independent reasons, people working on Proto-Yenesean have suggested u-vulers. Now it doesn't work perfectly, right? Like, why is it that the first syllable is rounded in all Chinese, but not in Proto-Yenesean and whatnot? So I, you know, I'm not, not a whole lot hangs on whether you're convinced by this, but it's kind of anecdotal evidence that this u-vular hypothesis is onto something. And then similarly, there's an old Turkic word, which is also a military title, Tarkan with a u-vular stop that was borrowed into Chinese as you see it. So maybe this is evidence that the u-vular hypothesis is onto something. Okay, so now just summing up the u-vular proposal. So in type A syllables, you get what you see. So this becomes a glottal, and then this becomes, this one should have a line under it. Sorry about that. And so should this one. So those are the proposals for type A syllables. And here are the proposals for type B syllables. And this is that conditioned change I was talking about. So you get a split where the voiced labio-u-vular stop becomes a velar fricative in general, but before front vowels, it becomes a vowel. And that's like, there's nothing more normal in life than something paddle-izing before front vowels, right? So that makes sense. Okay, so that's the whole proposal or that's the whole presentation basically, where, you know, let's go back to early on, basically, our understanding that we got to hear is where is the understanding that we end with, which is what does the phonetic in a Chinese character tell the reader about the pronunciation of that Chinese character at the moment it was coined? It tells them that, or tells them what its place of articulation is for the initial. In this case, bilabial, it's saying this syllable starts with the bilabial and it tells you it's rhyme. In this case, ah is the rhyme. So this phonetic says things like pa, where things like pa can mean pa or ba or pa, yeah? And in order to believe this, we have had to add a lot more things to all Chinese. So laterals and voiceless resonance and uvulars. But, you know, that's, or I don't know whether you want to believe it or not. Not everyone does. But basically the upshot is you can specify for each phonetic what syllable type is implied by that phonetic. So this is where I jump to the end and I am working on finding a way of romanizing Chinese characters that reveals this. So I'm not sure I've been very clear about that. But in this case, we have, this is the Shesheng series, the phonetic series of all characters that have this as their phonetic, yeah? And then what this means is a syllable-like lay, right? So it's a syllable that starts with a lateral and has i as its rhyme. And then the first, the middle Chinese reading of this character, like not this as a phonetic, we have to distinguish it as a phonetic and it as its own character. So the reading of it as its own character in middle Chinese is ta, yeah? But, you know, we know it's a lateral because of these yas down here basically, right? Because we know that in a series that mixes dentals and yas, we reconstruct the lateral. I sort of went through that, right? But now I'm just trying to show you this sort of thing I've been doing that probably looks totally perverse, so it's a kind of little dictionary if you like where I give the phonetic, then I give the syllable-type lay and then the reading of that character in middle Chinese. And then here's the next, this is the next character. So this character has two dimensions and then I write the semantic as a superscript and then the phonetic in Roman. So what this is, is this is telling you, okay, this morpheme has something to do with infirmity with sickness and is pronounced something like lay, yeah? And then in order to know, you know, it's middle Chinese pronunciation, you just have to look it up, right? But these are its two middle Chinese pronunciations. Now together, those two pieces of information allow you to reconstruct it in all Chinese, but I'm not doing that here, right? I'm giving you the pieces so that you can do it. So in particular, you need to reconstruct a voiceless lateral in order to get these readings, yeah? But the fact that this morpheme had a voiceless lateral in all Chinese is in no way indicated by the Chinese character. The thing that's indicated by the Chinese character is that it starts with some kind of lateral. So anyhow, this is my, you know, attempt to sort of make explicit the phonetic information that's actually contained inside the Chinese character and I've also sort of indicated to you how we go about figuring that out, yeah? So basically, you assemble all the middle Chinese readings of all of the characters in that phonetic, then you use this sort of Sheshan hypothesis, you use the Sheshan hypothesis as a principle to decide what it is that the common denominator of all those characters are, then you can write that down. And then in this case, it means all of these characters start with some kind of lateral and then they have I as their rhyme, yeah? And I don't know, you know, what the upshot is pedagogically, but like I think that it's maybe, you know, it's important for students to know maybe that the principle that you see at work in the example horse and mother is in fact at work in effectively all Chinese characters. Yeah. And the fact that it doesn't work in Putanhua is, you know, so much the loss for Putanhua. Yeah. And that, you know, maybe some kind of transcription like this could be a useful way to memorize Chinese characters, I don't know, but it might just be, you know, a waste of time because it would be telling you how they were pronounced in the Joe dynasty. But that information is, that's another point there to say that information about how they were pronounced in the Joe dynasty is like every time you type a character in a modern font on your computer, you're putting that information in there, right? It's there in the character. Okay, so that's where I will just, you know, go to the last slide where I thank you and then I'll stop sharing my screen and that's my presentation. Thank you so much.