 And so I want to make clear from what position I'm speaking. So this is not just Kapthatsu or Benewalensia, as we'd say in German, but I'm not a linguist. And also I want to say thank you. I feel really privileged or but definitely happy to be in such a circle because I usually don't get to hear this kind of discussion. I understand some part of it and I really profit from this very much. And so I want to speak more about writing practices. And then my second disclaimer is I should also admit that I am not a trained paleographer in the sense of ever having had a teacher in this area and so forth. But I got interested. So my idea is mainly in the formational history of texts, of early Chinese texts. And I've done this for a long time just with transmitted texts where you can do this as well just by comparing and historical criticism and so forth. But then of course went into this new field of manuscripts. And so what I want to contribute today also is maybe a third disclaimer. Isn't any new research. So some of you know all my points, probably almost all of them that I've made earlier somewhere. But I've made them in different contexts. And I just want to give some examples that seem to me to fit into this discussion of how we use a phonology. Sorry, how we use a paleographic sources for phonology, for the reconstruction or whatever words you prefer of whole Chinese. And they're not completely random, but they're not all very neatly related. So this paper is not trying to be a systematic discussion of something that I've published in this way. So I just want to go through a few examples. And now maybe my last disclaimer if that is one. I'm sorry I have something in writing here because I know what happens when I just go through my examples and then I start to ramble and go into too much detail. So I will, for the first part, be mostly following a written script here because otherwise it's a time problem. So let me start by saying that I consider it one of the strong points of the recent book, Back to the Saga, to include very strongly paleographic sources as evidence. So what I want to present is just something where I make this non-training that I just described productive in the sense of being unencumbered by a lot of theories of how paleography should work or the linguistic side of it. And just try to apply what seems to me to be fairly sane reasoning. And consider certain possibilities that I want to suggest how writing could have taken place and maybe even how the reading of these materials could have taken place and what that means to us. So a point with which I agree very much is the point you make in your book that the manuscript characters, especially from pre-imperial times, reflect sound changes more flexibly than the more stable orthography of later periods, which makes these sound changes less visible or maybe entirely invisible. You could, of course, then say in the 20th century, orthographic reform does something else, but that's obviously not what we're discussing. So this flexibility of early orthography really makes early manuscripts a very important new source for the reconstruction of old Chinese. And I think that is what the book makes abundantly clear with multiple really excellent examples. I'm not going there. But what I want to discuss today is that the lower degree of orthographic stability also creates problems with regard to extracting phonological information from the characters. I want to talk about a few aspects of early Chinese writing practice that I believe should be considered as possibilities. They're not necessarily really firm assumptions that I have. And I think that will become clear when I get to the examples. So I don't know if my title limitations to the source value is particularly good, but let's just present what I really want to talk about it and then see if it fits. So let's first come to what I think are excellent examples in your book of how sound change is reflected in different orthographies. And just to save time, I won't say anything about it. They just convinced me the evidence I know. And I want to come to another point where I think that not all changes of, sorry, changing or varying phonophorics in characters are as clear cut. So, and you've already prepared a lot of ground here. And I want to go to the example of Warring States characters used to write the words sung, bereavement, mourning or sung to lose and show that changes in phonophoric character components may be unrelated to sound changes and occur for more complicated reasons. And getting phonetic information for manuscript characters involves just to be a bit more systematic. I think three major activities. First is to identify the character components. Second, to identify the functions of these components that I would roughly distinguish in either phonetic or semantic or distinctive that is usually disambiguating. So some components really plain or semantic or phonetic role. And third, if that function is phonetic then to determine what this phonetic value is. So let me begin with this example that I just mentioned. So according to Hueshen, the character is a combination of the phonetic wang written under the significant ku to cry whale. Excavated manuscripts and inscriptions however do not confirm this. So I'm showing another character here because this will turn out to be the actual phonetic. So these excavated inscriptions and manuscripts do not confirm this. In fact, the word was written in Oracle bone inscriptions with the character sang for Malvary tree to which several distinctive ko were added to show it's not the Malvary tree that the character stands for. Then beginning in early Joe inscriptions the character simplified by reducing the complexity of the upper part of the tree graph to a form that now typically looks like the graph for elbows that we're back there combined with a varying number of ko elements. This makes the resemblance to the character for Malvary tree less obvious and may have affected recognition of the phonetic value. To balance this, I believe new graphs were incorporated into the character. Bronze inscriptions show the graph wang merged with the upper part of the tree character or just the tree. I don't want to be too precise about what is the upper or lower part of the tree. And the presence of this phonetic is understood as evidence of a labial initial M in Baxter 92. And I think continued in your book if I understand that right. And that's really where I have my question if there's another way of looking at the paleographic evidence. Which is a bit complicated. Since the earliest characters for sang do not have the phonetic wang and many of the later forms also do not have this element either. So let me go to this. Some of the bronze forms vary the original distinctive ko and instead use a foot or a combination of foot and sky corresponding to modern character zoe which makes some sense as a semantic component but not phonetic. This already raises the question of whether the function of wang in the other character forms is purely phonetic or maybe semantic or both. According to my somewhat cursory survey even the majority of jungle forms here they come, do not have the component wang but augment the abbreviated mulberry tree graph at the bottom but either the tree component or the character for the word si to die. So obviously it's least in the case of si here nothing phonetic and sometimes the right part of the si may be missing and I think that's what in your books reflected as it could be die. Just the left part of si. Mul interestingly of course it's not the phonetic value of mul that plays a role but I would still consider the addition of the mul component as phonetic in function because it reinforces the recognition of sang the original phonetic and then some have wang and some have a combination. Okay, let me see where I was. So again the oldest character forms for the word sang to lose. Wait a minute, I have another one by the way I should show this one because this is also mentioned because that is what paleographer sort of lists here at least some and I don't know where in the backstages of our book the information was from has actually mung with its outs at all but I don't believe that it's at the left edge of the bamboo slip and it's actually the same as this one only there wasn't space to turn the elbow. I believe that explains more successfully what the identity of these graphs are mainly it's again just the elbow with merged with a wang character and yeah and interestingly this is just the one manuscript that I worked on a bit more in depth in Jufumu and here you can see where and I've really given this some thought where I don't believe there's a change of scribal hand and you see sometimes you see a development of how a character is written there's a very nice example from Gordian where they say in manuscript three successive slips they have they use four co-elements and the next one sang uses three and then the next one only two so and this is of course something we look at in writing practice you abbreviate because you know there's no point in writing all these forms apparently and here there's no such developments and I find this significant so you can see they write this form with the full tree here recognized and helping us recognize the sound then they switch to the elbow maybe with the wang element in it I cannot be sure and then the si below but then they switch back to the tree and so forth that doesn't really seem to be a particular rationale behind this it's just what a scribe does and I think that's a significant point for me to emphasize that if we don't stay aware of this flexibility of that someone just does something to make it work somehow we might sometimes overinterpret certain graphic forms that we look at okay to sum up all these character forms of song to lose do not contain one Django forms predominantly rely on the mulberry tree part of the character is phonophoric that is a syllable that has no labial initial this raises the question of whether the use of wang in some characters was not was maybe for its semantic value just like the graph si and so and other forms if so then the very assumption that wang is phonophoric in the character for song to lose maybe misunderstanding that occurred at a later time closer to that of Xu Shen when the mulberry tree component went entirely unrecognized when its abbreviated form was mistaken for cool to cry wang was then assumed to be the phonetic once this understanding of the graph had taken hold this would have been motivation enough to discontinue the use of the tree component and to favor the wang graph over city to die which did not have so I mean the advantage of a phonetic connection in such a circumstance it seems possible to me that the word song song never had a labial initial unless there is another type of evidence for that sound that I'm not aware of this is really where my competence just leaves me so purely from the paleographic evidence I'd say it's unsure I've used this example to demonstrate that the identification of the phonophore it can be difficult as first the writing system itself was not yet stabilized that is alternative character forms coexisted that allow different interpretations of the function of character components and make correct identification of the phonophoric problematic second even where the writing system clearly prescribed or at least favored a certain character for a certain word the way the graphs were written in practice created ambiguities which again make the identification of the phonophoric difficult and the third question is worth asking how consistently did the writers I'm avoiding scribe here and we tried script for a long time it's just not really practical so let me just say the writer not in the sense of writer of another the writers and readers in early China how consistently did they associate certain phonetic values with certain graphs how important was this identification to their reading practice and what guided ascribes or copies actual representation of a world so here I want to go to another case that I find interesting the sentence final particle and Shin and Han manuscripts and what I have to say about this is based on Unishi Katsuya's work was shown that the two characters indeed stand for two different I'll be a synonymous words II and year rather than both for the word year as earlier transcriptions suggest so most of you may know that you found this character and then people give a year behind that in the 70s modern day transcriptions and so forth but of course they're not claiming anything that just not being clear they're not saying we assume it stands for the word year but it certainly creates this impression maybe they just want to give us a semantic hint or so but anyway it's it's a fact for the users of these transcriptions so Unishi identifies the particle E as a feature of the language of tin corresponding to the particle year that was used in the East and in true his study of the distribution of the characters shows that after the establishment of the empire the tin tried to enforce the use of the E particle especially in administrative documents and at the end of tin and early on it begins to disappear again a manuscript that I've studied in some detail the so-called Laozi B manuscript from Ma Longdui starts out using the particle E uniformly for the first 13 times that such a sentence final particle is used and all these instances occur in the first 17 columns of a manuscript of all together 252 columns until this point not a single year appears and in the same column 17 then after three instances of E the use of the particle year begins and continues without a single recurrence of E for the rest of the manuscripts all together about 350 times so the function of both particles is clear enough they appear to be completely synonymous but phonetically completely unrelated going by the writing system we clearly have to read different final particles within the same passage of the same text in German where it changes written by the same scribe who commanded an excellent calligraphic skill the interesting question to me is to what extent this scribe considered the two characters as exchangeable did he grudgingly write the cumbersome E with many strokes because it was required to but would nevertheless have pronounced the sentences ending in year in his region soon enough the scribe gave up and wrote the character which he was apparently used to namely year and which was also quicker to write likewise would a reader of this passage at the time of very similar sentences have read E first and then year or would most people have read such an obvious function what the way they were used to whatever the character was for example a reader who believed that year sorry E represented a higher level of style or political correctness could have pronounced the manuscript's year characters or the many of them as E or true patriot I'm making this up it's just speculation here could have pronounced even the E characters as what to him was the proper final particle so he would have read year just like scholars who transcribed in the 20th century they usually always read year because that's what they believed it really is so and I think that has such speculations which they are have a bearing on maybe how we look at this evidence so a similar question arises when we look at characters and true manuscripts that undoubtedly stand for the numeral one when we read these manuscripts today there's little else we can do then pronounce each of them as E in modern Chinese as we used to probably assuming that this is the word written in the manuscript text and then of course the dental final predecessor of today's E in manuscripts from different true tombs and in very different texts a character consisting of e feathers of a nun able is used for the numeral one it is very unlikely that this character phonetically represents the word E the character suggests rather word compatible with the phonetic nun and indeed it could be cognate of a cognate of a Thai word for one that has a Vila nasal final and the tip for this article that I'm citing here I got from Wolfgang so thanks again and also another interesting one that tries to sort of brush this under the carpet and make it a Chinese word again but I haven't included this information because I found it a silly article so in Thai essentially this complex character is used in the same passage as the single stroke one which in this manuscript occurs only as part of the name Thai so when the single stroke so it seems possible that the manuscript uses a local word for one that was a Thai cognate considering that the one stroke character does not have a phonetic value as such but could be could be understood just as a digit used in alphabetical writing system it is possible that readers of the manuscript would pronounce both characters for one the same way with a nasal final but it is also possible that they read the manuscript in a more diglossic way following a convention in which as part of the name Thai the word one would be pronounced with a dental final and otherwise the word for one would be used just like we use different words today for a first lady I put this here and on the one hand but in the school we call the principle and no one has a problem with this basically both means first so there's another complicated case of variation between characters for a word or words meaning one here we go other manuscripts such as Hang Xian used both the one stroke character plus a combination of this stroke with the character for guh, halberd despite its clear and consistent structure transcriptions of manuscript text identify the component of the character as e an arrow although it has an additional stroke and it's not really this e if we look at the structure of this component most probably because the combination is listed in the shuwen as the guhwen form for the one stroke character as neither the halberd nor the arrow are possible phonetics for the Chinese word for one the word written with this unfamiliar character for one may have been pronounced differently in different local languages this leaves me uncertain about the reasons why true manuscripts have different characters for one a phonetic difference is a possibility but a semantic differentiation is just as possible as is a combination of both or least likely neither if the different characters did not indicate any phonetic or semantic distinction they could also go back to locally different orthographic conventions the likelihood of such a scenario is usually reflected in a certain distribution of these variants they would coincide with different hands or occur where the same hand copied texts from different sources in the example from Tai Sheng Shui or Heng Xian neither seems to be the case but I have one example where I think I can show different source texts as the reason the so-called Mawangdui Laozi a manuscript silk stroke containing several texts copied on the same silk manuscript actually everything after three is a bit uncertain because there are no designations of texts in the manuscript so it's just based on logic and how we read it and the names are made up anyway they're not original and I forgot my own convention of putting an asterisk in front of them to indicate that but the other texts are I mean there are graphic distinctions but more than six so there are what we could call paragraph breaks today but in any case without going into too much detail what is completely unproblematic is the division of these first three and the first one is the canonical part of the guidelines part and then the explanation so these are clear differences and in these manuscripts there are several orthographies of words that change just when the texts change so I think this is it's reasonable to seek the reason in someone copying from sources that come from different places or follow different orthographies so the example of the word Song shows the precursor of the modern standard form as well as an obsolete character whose phonetic swan is close enough I think to be a mere orthographic variant reflecting no phonetic difference or at most a slight dialect variant but a case where I see the possibility of a phonetic difference is the variation between the full form of the character for hope and an alternative forms that were used consistently across large geographic areas and for a long span of time from the warring states into the early empire while the word is written in its full form whenever it is used contrastively as the opposite of Qian or of Qian in the frequent on the other hand then in the frequent collocations run ho and are ho it is written with an alternative character that required fewer strokes so typically this character is the one for go hook with or without an additional bamboo or grass component these collocations occur more frequently in texts than the word in its full contrastive sense does so we could assume maybe that's why they wrote it in a shorter form because it's just a curse to offer but still thank you they do this in a very systematic fashion so it could be a mere matter of economy of writing but it is still possible that an economy of speaking is involved as well I don't know if I should use such words among linguists here but as the word is also pronounced I think it could be the word is pronounced in an unstressed or contracted form but I don't know how this would play out in a linguistic sense I just want to mention the possibility because consistently this different form occurs in collocations of two particular syllables but I think it will take much more accumulated evidence for understanding early writing practices to answer this kind of question at this point I see from the paleographic angle we just need to do a lot of collecting and also sort of coming up with some speculations that seem possible and then just collecting these as well without making a theory of it and being very sure that is what answers the question so somewhat easier to understand phenomenon than the alternation of different characters is their abbreviation which brings us back full circle to the case of Sang in the characters for Sang the phonetic was only abbreviated and when it became unrecognizable it was augmented by semantic or phonetic components that help recognition but there are cases of complete omission of the phonophore component or its simplification beyond recognition so Shen is a good example I think it was frequently written without the phonetic element and there are lots of examples for this in other cases omission or just abbreviation of the graphic component was indicated by an abbreviation mark two horizontal strokes that were integrated into the space of the character not placed after the character that's an important distinction because these strokes here you can see these strokes that I mean occur in the character space here it's not quite so clear and as opposed to in the position where the red dot is here now this position you would find the ligature or repetition mark so these need to be distinguished but I think this is much too little still much too little discussed phenomenon and it was discovered sometime in the 80s I think by Lin Xuqing and I think a doctoral dissertation so it's the knowledge of it has been around a lot but it seems to me that people keep it a bit to themselves and don't bring it into the discussion prominently enough for those people who do not do the polygraphic work to recognize this possibility so the characters for Chiang are typically written as Gongbo plus Ko plus two horizontal strokes and sometimes another Li component clearly semantic strength judging from the overwhelming manuscript evidence if we were to overlook then not yet widely recognized function of these abbreviation marks we could be misled into assuming that the left-hand part of the character is the only possible phonetic in this the homorganic initial and identical final would support such a misconception Xuqing was apparently already looking at a form corresponding with the modern one and isolated the insect at the bottom right I'm sorry so this part here and then considered Gong as the phonetic the abbreviation mark indicates a difference between the character as it is written and an underlying standard in the writing system and that is important to me I think this is really a very big question that we always need to stay aware of that I'm not denying the existence of an idea of what components a character should have ideally I think there was an underlying system somehow just like we have our alphabet but then if we look at our actual handwriting it differs significantly in many places only it never creates not never it does create it doesn't necessarily create a very big problem for readers because there's so much context just like we often overestimate the extent to which we actually hear all the individual words of the partner conversation and that's why we understand what they're saying and we were not aware of how overwhelmingly big the role of context is in how this conversation takes place so this difference here is indicated by the abbreviation mark between the character as it is written and an underlying standard we either need to imagine an absent component or recognize a present component as a substitute of another more complex one in this case the co yeah is marked as an abbreviation stands for the actual phonetic Jiang and there are a few characters at least to confirm this underlying identity of what is mostly represented over the mouth but I think we're not always that fortunate to have these examples where the actual phonetic is still indicated to some extent and then we're really lost in how to interpret the characters and I have much more to show originally I wanted to make this talk mostly about the famous character for Ren I, Ren Eder, Ren and but then Wolfgang threatened this really extensive and complex article about that that says it all so I don't need to repeat this to some extent maybe if I can use a few seconds more for this there are questions that go beyond just this one example and maybe here I have Wolfgang's article at least I wanted to show this information and so the only aspect I want to point out now is that graphic components in these characters are easy to confuse in some cases so a notorious one is are these three Ren and Shi and Gong usually Gong is the easiest to differentiate because often there's an additional little horizontal stroke and the Gong should be the most curved of the three and then Shi somewhere in the middle with just one noticeable curve and then Ren not much at all but in practice in most manuscripts these distinctions are very very blurred so the one example for Chiang where it should be clearly a Gong that looks like a Danrenpang absolutely and then in the others you can see they do a slightly better job and then I just collected some examples I just copied out of my own book here is where we have this is for Chi Chi Chi threatened this way so you have lucky it is the phonetic it seems the Shi so that is then the real identity if you will of this form and then all the rest is missing I mean it's almost the entire character is missing but at least the phonetic is there and then in in this one same thing the whole bird is missing and then you have this Shi here this stands for Nian so it seems to have Ren as the real phonetic and here the general assumption was I don't know if it's just mine because I went by the modern graph that it has to be Gong so that's why I still this Gong here but I learned from your book to that Shi is really phonetically compatible so I think no isn't it? oh no? no sorry with me I'm here in this yeah so that and I just looked at the modern character possibly I don't know what I did at the time and thought oh this is a Gong so that's the identity now of the old graph but maybe I just didn't get this point so it we should understand it in its old identity as Shi in this one anyway so there's a whole world of possibility that this opens up and another thing that bothers me a lot that I think and then I think I'll wrap it up it's still very prominent in the discussion that is Hushen's practice if I'm familiar enough with it of always splitting the character in two so there's this mantic part and then the whole other half is given as the phonetic and then you have to go there in the book and find again that this is a more complex one and only part of this is really the phonetic the underlying one and I think that blurs a lot of distinctions in how we read old graphs that we don't go right away to the real core phonetic element if you will and then this Ren Yide Ren written with Shen Yide, Shen and Shinse Pan is really a very good example because the Ren seems to be the phonetic and then when Hushen gives you Qian as a phonetic I think there's no way from the old characters how you can actually distinguish this and I didn't do a very good job to find the best examples for reasons of time but here clearly we would let's say these ones see the full Shen Yide Shen on top but sometimes it's just written in such a contracted way that it looks like only a horizontal stroke and then it looks like Qian so we cannot go only by what we identify purely graphically there is this element of interpretation of this and thinking about a possible scribe of practice and where they do a sloppy job I just won't be bothered and in any way it's neither Shen nor Qian that is the phonetic it's in any case Ren that is contained in these right so I think there's a lot of noise if you will of data if we worry about Shen and Qian and well it's all the time just the Ren component I think this is these other examples were just meant to hear how similar abbreviations occur just purely graphically and this one is just another notorious case that I wanted to mention where the component that is like the modern Si without the call and then modern Tai are really not easy to figure out and the newest transcriptions of manuscripts that want to be more precise sometimes then write both they give you the Si and the Tai just to be on the same side but I'm not sure they're both there it's just they resemble each other and they're just very gradually different forms of writing what is sort of the same thing but I guess I promise three times to shut up I think that's it okay thank you