 I will basically delegate most of the time to a discussion of the problems one sees working with true texts and adopting a method of chronological reconstructions. The way they're really dealt by people that they will try to make sense out of texts, when they need to translate them. That's my experience. I basically translated the Bordeon and the Chambord texts completely. And I had just to deal with the kind of choices that are made to make readable texts. How do you choose between one strategy and another? How do you see what the prevailing strategies in the fields are, which are not always strategies that conform to best phonological candidates, best phonological matches? So mainly I deal with these points in a sequence. What I believe are the main features of the book, of the new system. How they deal with new magic sourcing, but especially how scholars basically approach these sources. What are the common tools they use? What are the common problems met by scholars? Then I will focus specifically on the usage of Danyvan's dictionary in the book as a way to test some hypotheses and to see some counter example in the spirit of very explicit in the book of falsifier big, I mean really making claims that can be tested one way or another. And specifically we look at some detail, three examples. And then I will look at some other examples relating to relative problems from excavating text and it will offer a quick conclusion. Now, we were three of us from the University of Washington and we were trained with this book. We were trained with that book because we were weighing into old Chinese. But in a way compared to the approach taken by the 2014 books, it's a traditional approach in the sense that if you compare it with the testability of the claims and the structure of the hypothesis, the new book has certainly bolder hypothesis from a point of view of really epistemology. But it's not only because of epistemology. It is mostly in the good part, at least, because of new sources and new sources of material, Hmong and Taikada, I think, that are brought into the system and also much more recourse to internal reconstruction, morphology of old Chinese prefixes, and so on. At the same time, there is this issue of the hypothetical-tutative approach. And I'm intrigued by this explicitly, this approach coming from the history of science is the way that you test things, just like in the Einstein example, very famous, that was tested by an eclipse between two claims. So hopefully, the Einstein claim is a claim that basically comes not because of a slow addition of new data that brings to some more generalized hypothesis, it comes from some conceptual breakthrough that then needs to be tested. But when you test it, you find out how it compares with the established method. At the same time, looking specifically, and just limit myself to the script, the true script, looking specifically at some usages of this, of data from the script into the book, into the system. Sometimes there are issues of falsifiability. Can really these claims be tested? To which extent can they be tested? Obviously, even if they are only partially tested, but the way they can be tested can improve. This is also part of a normal way you build this hypothetical-tutative system. Now I want very briefly to go through the three main corpora we have for true script. Most of us are very familiar with it. A few, maybe, are not. And this, anyway, allows me to go back to the fact that the data that comes out of that constitute true script that are in valueless dictionary of phonetic laws come from single texts. And these types of different features. For example, the DeGordian, which are basically the first that really wrote us into an extended usage of true bamboo manuscripts. It comes from an archeological exhibition. We are sure of the source. It has been studied for more than 20 years. And this translation is very simply annotated by Scott Crook. Basically, it's really a point of reference. The way we treat the data coming from Podia are not the same as the way we treat data coming from the Shanghai Museum manuscript. The script is very, very similar to being related. And now, basically, you don't do, when you look at a graph in the Podia that has yet to be puzzled to be solved, you immediately use data from the Shanghai manuscripts. I mean, the two things are interrelated. They're just one field. And I particularly found that familiar with this, because I went through the trouble of trying to create a complete translation of that. I mean, some of them are broken in the order. It's unclear. Some of them are more incomplete than others. Overall, most people would agree that they reflect just one kind of script. Some texts overlap. We have the same texts in Podia and in Shanghai. At the same time, the problems are more extended. Sometimes, when we look at one passage in one of these texts, we need to be aware that, while the text could be really incomplete, there are even issues if the text could be forged. And then, generally speaking, these texts are accepted as being authentic. But this is not the same as having texts that come from two inner archaeological settings. The third corpus that is coming into full fruition but is still just been published year by year. We're going forward, I believe, by the Tsinghua Ministry. Again, we are not sure about the provenance of this. I would like very briefly, basically, we will not really go into this passage. We will go into the table that comes from it. This is a translation by Cook of two passages one from the author. I tried, basically, to do an investigation of how many of these characters are, basically, can be taken as standard version of the, basically, equal to the standard script of the Tsinghua Unification. And which one are either Jiajie or just conjectures. So if you count it, you would, in fact, more or less contain around 12,000 characters. Now, I did a sample of around 1,500, so I could control it like that. Let's say that out of 1,500 around 1,000 just stand for what is called then. So the same character. You don't need to have any interpretation, even for the conversation series. They're just the same. And instead, all the ones that are followed by something into brackets are around 1,500 out of 1,500. Now, most of them, let's say 70% of this 500, 70% of this character that require anyway a kind of re-transcription. They need to be interpreted in terms of the standard script. 70% are trivial. I mean, they're basically at the same phonetic component, the same transition series. Around 20% are less trivial. But in most cases, we can get the phonological reason from going to one character to the other. Now, we use character in a pretty informal way as representing a word that we not always have the full thing. But obviously, there are graphs that stands for the word that is normally written now with the standard graph, for example. Now, in some cases, this kind of equivalence could be basically due to some special phonological environments. For example, this one with a glottal soft correspondence to the initial nasal, it often comes to the kind of sound environment when a preceding word is followed, is ending in N, which basically also reminds us of the fact that it's not really the two characters that stand for each other in general. I mean, in a single text, there is a single instance countable of a certain character standing for what would turn out to be a different character. And these are the problematic ones. They are around 10% of this one third that are in general not fancy. And this one, in some cases, is just conjectural. I mean, they're just seen from the context that they're unknown, or they can be broken into known parts. Sometimes the fact that they're broken into known parts doesn't help. They're still basically unknown entities. But they often decide, for example, this Schrodinger, they decide books that have a related version from that matching word or character in that environment. And we assume that this unknown thing should match with a known character in that received passage, and sometimes just from the context. In other cases, it's basically just a very continuous, consistent way to write one character as another in running, in the magic, is written with a shun radical and a hand in the heart, which is also a phonological fun way. But this is kind of a, it's still a special case. It's still more complex than this. But at the same time, it's a straightforward equivalent. It holds both the time. Most of these others are really like blank boxes. We don't know them very well. The way we deal with these graphs is, first of all, we try to locate them into a dictionary of graphs. And this is still being the reference. But basically, the state of the field, the fact that the field is basically just really starting to give us tools, is clear from the fact that this dictionary has been conceived before all these discoveries, I mean, before these discoveries of people again. It doesn't really have data from people again, Shandong, and Qingguan, but it's still the best we have. But we integrate it with other things. We integrate it basically with data from databases. Academia, Scenicum, Wuhan, and Chinese U. You enter form, and they give you all the variant form. So obviously, it's according to their interpretation of what that form already is. You cannot just have an unknown character and try to find it out. You can just, through these searches, see how a given known character is represented in different texts, according to each editor. Each one of these is an editorial decision that is indeed this character. In many cases, this is trivial. In other cases, it's potentially problematic. The source I'm focusing here is Bayerian's dictionary. I did it because it allows me a very small skill to test some claims put forward in the book, which use Bayerian in sixth and seventh occasions is never for very major things. I'm not trying to have an overall outlook on the reconstruction as such, just to see, basically, two things. How the two-spread is used to build this theory, and eventually, if this system can lead to better practice in ideography, it can help ideographers to build the text. Now, the version of the text cited in the book 2008, this is the one, the latest, 2012, is three times larger. Partly, it reflects the fact that, really, there is a normal amount of work going on in this field. But partly, also, basically, it makes more explicit what is already here. And here, it's really called the dictionary. It is really a collection of glosses. Bayerian is really, basically, a collection of glosses. Meaning, it's just a problematic character or more as problematic as it would be treated if the sources were not manuscript sources in a normal JGF dictionary are treated by Bayerian in this dictionary because they come from manuscript sources. And at the same time, quite clearly, each one of them reflects a specific editor's view on a specific problem. We look, basically, at three claims. Three claims are made in the book and kind of test how the usage of Bayerian is related to this issue of testing the system. Because the fact that the claims are based on this version and there is an expanded version that is basically 250% more data, then basically allows you to check how things have been going in the past only a few years, but just the material is much more. So there are three claims that I'm looking at. And one of them is about what I call split of gong and gong, which means I kind of use the terminology that Baxter and I do use, basically, when he was looking at evidence of having one rhyme category that was traditionally treated as a single one, split it up into two. If, basically, there is no inter-rime in between the subset and then you 3D view it, split the category. In a certain sense, this is something similar, meaning these two words have the same middle Chinese transcription. And they were, in general, reconstructed also with the same old Chinese reconstruction. So one of the rationales for this different reconstruction here, one with a meter and the other with a eubola, is that in BYL by Yunan, there are pages and pages listing words that are in JG connection with gong and in JG connection with gong, and the two series don't overlap, which basically means that, at least in the mainstream in the book, the two sets really seem to be more than just separated by chance. They seem to be a systematic way in which the sets are distinguished. Plus, gong reconstructed with an eubola serves as phonetic for characters that are also reconstructed with an eubola. So it's consistent. It's basically an hypothesis is made. Then it is defended on the basis of new evidence from the manuscripts. And then additional reasons are given. Now, in BYL by 2012, there is one counter example. The counter example is slightly indirect. It's not exactly from gong, but it's from gong, which is basically still with a veller and which interchange with an eubola here. So gong writes gong. We saw first that one was reconstructed with an eubola. According to the pattern, we think should not have interchanged with a veller. What I want to point out, here I have my small advertisement of my own translation of the passage, but it's also that it's basically one solution to one specific textual problem. The graph is this. At least this is the way it's transcribed. There is gong here, there is teo, and then there is the vessel part. So here. So these three parts. So it seems quite clear that the phonetic is gong. Sort of here. Now the story is about a dupe that basically has understand how things really are. She's a good ruler. And a younger ruler that is supposed to take over, but doesn't understand at all how things are in the world. So the story is that he goes around in the countryside, looks at peasants' family, and look at how you cook pickled sauce. And he understands that you need to do it in a special way. He knows how to do even this kind of very small minutiae. So he's a good ruler. His son doesn't even know how to recognize him in a field. Doesn't have a clue. So that's the way you kind of understand what is going on in the text. And this is the point where we make the decision where this gong. And we decide it should be earth and jar. So it should be an earth and jar that should not be covered in the process of cooking, where does this come from? It comes from Chentier, which if you know how it precedes in maximum value, you're not too careful about the chronological details in most cases. But this is, at the same time, a typical entry from Ba Yulan. This is an entry. And what I want to point out in the spirit of this falsifiability is that basically when Ba Yulan has his five pages of burrows from gong, and then for the other, gong, and then Ba Yulan, his sources for putting together all this list is basically textual glosses, like the one Chentier put there in this case. So there are two different issues. The fact that if Chentier is justified here in having that glossy, which would consider this and stay for just phonologically unsung, and the fact that in the book, the fact that gong and gong consistently do not intertwine in this two-ji-a-ji-a system is considered a fact. It's not problematized as a set of hypotheses of different scholars taken and put together by Ba Yulan. It's basically the fact of the language. And I'm basically emphasizing that often there are actual reasons to choose one reading against another. And these reasons might not be controllable, but at the same time, they end up in the pool of data that is treated as a fact. So sometimes there are issues of falsifiability, meaning the data like this is just going to be taken out as problematic. But then what about all the other data? They've been tested the same way to check that we were not conditioned by similar contextual reason or with possibly problematic phonological reasons. So this is one example. The other example is actually a kind of confirmation. It is not a counter example. And there are two series, Yang and Yang, similar powerful and significant changes. Reconstructed with an eubolier for Yang because of connection with the series, eubolier items. This other item has connection that basically lead to reconstructed with the liquid, because in itself, there is a certain consensus that you reconstruct middle Chinese yam as we're having these two sources. At the same time, Bayou now, again, neatly provides these two known overlapping sets. The new addition has one exception, but this is not a real exception because we know the two merged. They only merged after the bamboo manuscript. So this manuscript, Bayou now is not only bamboo manuscript, this is Han Dynasty manuscript. So the only example when the two interchange is from the Han Dynasty. This is actually something that comes as, let's say, testable new evidence, and the evidence is in favor of what is represented in the form. So it can go both ways. Wu and Wu, also, Hong Kong's middle Chinese, reconstructed in different ways. Now, one of the reasons to reconstruct as having eubolier is that there are context, there are seething context, and there are word-families relationship with eubolier. So this is, again, the hypothesis. On this kind of hypothesis, we just think that, sure, the two series should be separated. But here, the counter-examples are very numerous, actually. In fact, they are treated in a separate note in the book because it's the case where, really, counter-examples are in large number. Now, what is considered a genuine counter-example in the note, meaning a counter-example that is simply not explained the way. It's simply recognized as an exception is this one. This one, because there is a matching passage, I think I'm shooting one, that basically makes extremely unlikely that, here, we should have anything else by eub resistance. So this is recognized. But this one is cited and not recognized. And Bai Lan gives this gloss. But it is possible to give this other gloss. Now, how is it possible? This table will basically give an idea of why I consider the system. It is not out of control, but it is a system that is still very problematic to deal with. The alternative explanation of this word, SPM, which obviously cannot be on phonological ground, but it's not on phonological SPM. Is this graphic element here? Is one of the graphic elements perhaps the most ubiquitous mysterious graphic elements in the program? Because it really seems to serve for, well, widely different. It stands for all this in different contexts. And so it's a kind of jolly element that can serve different purposes. And we probably just didn't crack it the right way. Probably there are at least two subsets. But what I mean is that the alternative hypothesis of having biennium here comes from the fact that the graph involved is this graph. So, well, this unknown basic polygraph that can stand for so many other graphs, which also means it's perhaps an easy way to get away from a textual problem. If equally, it would have been equally possible to choose one of the other candidates. And at this point, how do we choose we should have words and debates? Well, in good part, from the context and plus on the polygraphical ground. This is a dialogue between Cautius and Zixia. Zixia is saying basically that the words Confucian are beautiful. Perhaps it's preferable to keep words instead of debate there. But again, it's a textual. I have one more example. This example is based on a graph. Again, that is quite interesting. It's interesting because quite clearly it doesn't match with the phonology. It would be the n-final gene would not serve for something that is pronounced with ing. So it has been surmised for a long time. But actually, if I think it's ing, furthermore, ing and ing these are just very important. But they are basically almost interchangeable. So here I might seem a bit fussy in going over these minor details. But I'm basically just trying to see what can be justified, what can be tested, what is very difficult to test in this subset. So this is straight from our text. And it says that also taking in good consideration the fact that gene and ing are related. So perhaps we should actually split gene or Chinese cream into its two meanings. Because we know it has to be one is both small and the other one is pretty. So the proposal here is that probably the both small meaning come from ing, to come back. And the pretty meaning come from ring, liang, to be compassionate. Now, this is all possible. But how do we test something like this? Obviously, sometimes if you don't make even, I mean, the issue maybe is if you should have this in a note or in the text. I mean, it's like in a sign. It's a possible way to think about some future problem. At the same time, it's all presented in a text that is will be a reference for years and present itself as a kind of careful, hypothetical, deductive text. So I'm just taking some small issues with this, just showing some inconsistencies that we have with the two strings, particularly, because it's a dead end thing. Seeing a thing is this. Joe is general, recognized by everybody, but is generally recognized as being the original character standing for an order. So how do we reconcile the different initials, T and K? Now, it is difficult. Obviously, there are two kind of supporting pieces of evidence. One is that there is a kind of a couplet. I mean, there are two cognitive words in one in Tibet and one in their own related, where basically we see that there is a, I said we see that with prefix or what we in Chinese will consider all of this. So there is a toe crew that could give us the crew, the complex image. And also, there is a, again, an hypothesis that comes for internal reconstruction that the T prefix could account for this. Again, I am wondering also because I have a special order in sign of Tibetan or Tibetan or Permanent. But how and how is having this prefix? I mean, it's not base only. I know it doesn't come only from that, from that relationship between Joe and Joe. But are there only five? It just seems to be, as a supporting evidence, if there are only five or six and one is problematic, almost 15, 20% of the evidence is problematic. The last example, I'll be a bit out of it, more or less I'm dead, comes from an example in the Shambu that I, something that we bound into. So this is a place name. Place names are particularly unreliable. You never know exactly how they're written. But from the context, it's in the only way we can make sense of this, which is skin, is that a certain king has lands in a certain territory. It's not that he, obviously initially people, the Boombu script are such that sometimes people thought it was about a skin problem. I mean, sometimes you need to reconstruct the whole, sometimes you really understand the single passage based on a single word and go around it. But now that we have a decent understanding of the text, we have no doubt that this should be a place name. And once we know it's a place name, we know from the chronicle, this is a two king. We know which time it fits that he actually had this new territory counterpart. It fits for the rhyming, but what about the initial? I mean, the initial, people accept it because it seems to make the better sense in the context. But again, can we, how do we justify it? Maybe we cannot, or maybe we can. I mean, can we test it? Obviously, there is no prefix, p, but it's just another example of things that are problematic, but editors make these decisions. I mean, most of the time, the text we read, the text that by Yilan, we have as his entry, has this Georgian relationship. The conclusion is that the evidence from the Shoe Speakings is open problematic, that sometimes, at least in this case, it's not clear how you test certain hypothesis. And especially, it's not clear what is given as a normal, suitable way to test these hypothesis. Maybe there is no general suitable way. Also, as a basically practitioner in the field that uses reconstructions to make decisions about the idea, I'm still using Shoe Speakings, but I mean that one, because it's more systematized, but it's based on Western 92. I don't think it's yet, I mean, it's really a task for the future to make the old Chinese new reconstruction to make day-to-day decisions.