 So I'm glad that Mark did his presentation just before me because it makes it, he paves the way. And the question is that I don't have to delve so much into. Now, this paper has seven sections. And I'll name the sections as I move from one to the next. And there is an appendix at the end on pages 10 and 11. So we'll look at that in time. So like Mark's paper, our paper, the bills in mind, is a hypothesis on the origin of all Chinese pharyngealization because we don't think that pharyngealization was always there. So as Mark showed, as Mark said, we reconstruct two sets of pharyngealized consonants. As you can see at the top of page 2 in that chart. And it should now be obvious to you that the inventories of pharyngealized and non-pharyngealized consonants are parallel. And in our case, they are phonemic. So Mark also showed what the reasons are for reconstructing pharyngeal. So I won't go into that. One thing that's worth mentioning is that Norman 1994 regarded pharyngealization not exactly as a characteristic of initial consonants, but as a characteristic of entire syllables. And we think that it is not the case. So we took the liberty of modifying Norman's theory a bit by saying that pharyngealization was a feature of the initial consonants of the main syllables. And the reason we did that is because type A and type B words rhyme apparently freely in the O's. And if the vowels were pharyngealized, we would not expect them to rhyme. So we pushed pharyngealization left towards the beginning of the words, presumably in the onset. So that's why we have these two series. Now, as we observe on pages 73 and 74 of our book, this is typologically unusual. As Mark said, I'm very glad that he showed that it's not so unusual as it seems at first hand. But the point is that here, we think of typologically unusual features not as impossible characteristics, but as characteristics which when they arise do not last long and stable. They are transient. They tend to disintegrate fast. So the evidence we have for reconstructing pharyngeals dates back to really to the end of the old Chinese period or to the beginning of Han times. And so the evidence for pharyngealization can be put at that time. And we have no evidence at all on what pharyngealization, sorry, what type A and type B should be at earlier on. Let's now move to section 2, where we outline a hypothetical model of how such a contrast may have arisen. Our proposal is spelled out in table 2 at the top of page 3. Table 2 has two columns, type A on the left, type B on the right. And it's in three stages. The last stage is the old Chinese stage. The third stage is the old Chinese stage that we reconstruct. And stage 2 and stage 1 are earlier stages. So for type B, it's very simple. There's no distinction between. There's no difference between the three stages. The consonants remain the same all along. So barring, of course, all the changes. There's no non-pharyngealized consonants, we assume, were non-pharyngealized at every stage. In type A, however, it's a bit different. The terminal stage where the pharyngeal consonants arise, we take back to an earlier stage in which you did not have a pharyngealization was represented by an independent segment, a pharyngeal segment, a pharyngeal fricative, which is the ion of Arabic. So clusters, the difference at stage 2, the difference between type A and type B is that type A words begin in clusters of a plain consonant plus a pharyngeal fricative. And that, in turn, goes back to an even earlier stage, stage 1, where type A consists of a plain consonant followed by a vowel followed by the pharyngeal fricative followed by a second copy of the same vowel. And then a possible code. So basically, we suppose that the type stage A forms lose the first vowel, then the cluster is formed, and at stage 3, the pharyngeal consonant influences the initial. And they do a sparyngealized consonants, which may not last for very long. Now with this reconstruction in mind, let us move out of Chinese into Sino-Tibetan and into the branch of Sino-Tibetan known as kukicin, including the language lushai. The kukicin languages do not have a pharyngealization contrast, but they have another contrast which bisects the entire inventory, the entire lexicon, and that's a vowel length contrast. Some words have long vowels, others have short vowels. And there is a proposal by Stavastin in his dissertation, 1989, who claims that there exists a correlation between lushai long vowels, words with long vowels, and Chinese type A words, and lushai short vowels, and all Chinese type B words. There is cognate. When you have cognates between lushai and all Chinese, they tend to conform to these two correlations. Now with the hypothetical model that I have presented in table two in mind, let us look at figure one at the bottom of page three. Figure one has two branches, and the root at the top, the root is the ancestor of lushai and all Chinese, which we assume to be proto-synotypeton. So the right branch, which goes into all Chinese, is basically the same development shown in table two that we just saw. Now how does one go into lushai? Well, very simply by losing the pharyngeal consonant between the two copies of the vowel. So that gives you, in fact, a contrast in lushai between words with short vowels and words with long vowels. Very simple. At first sight, it may seem a little cumbersome to reconstruct the two things together, pharyngealization and vowel length. And of course, it's easy to derive both. But the point is that similar contrasts exist elsewhere in East Asia, specifically in proto-Austronesian and in proto-Austro-Asianic. Now let me first tell you about the Austronesian contrast. And that will be in section four on page four. Now Austronesian is interesting in that there is a strong preference for disillabic words. There are almost no, you cannot reconstruct, you can reconstruct almost no monosyllabic words to pro-Austronesian, unless they are function words. Content words are not, monosyllabic content words are not reconstructed. There are indeed meaning associated monosyllables in Austronesian that exist. And you can reconstruct them. But interestingly, they never occur as monosyllables, presumably, because there is something against monosyllables in that language. When they occur, it is either reduplicated or preceded by a dummy syllable or preceded by phonic material that cannot be made sense of. Now these, but they never occur as simple CVC words. Austronesianists have recognized these for a long time and they call them roots. You can find collections of roots in the words of Blust and John Wolf, for instance. For instance, one root, suck, s-e-k, s-shua-k, means crown, crowd, something like that. And incidentally, it seems connected with Chinese site. Now you cannot reconstruct the form. It's never a free form. You have suck, suck to crown, crowd in pro-Austronesian. In pro-domaleo-polynesian, which is a branch of Austronesian, you have has-suck to jam, crown, crowd in forms like bes-se-k, das-se-k, das-se-k with similar meanings. But you never have suck all by itself. Now these forms in which you have a root, these monosyllabic words where you have a root at the end and material that you cannot recognize at the beginning, are probably old compounds in which the first part of the compound has become unrecognizable, sort of like crown in cranberry, because of phonetic erosion. And this, in fact, suggests very strongly that pro-Austronesian was stress-final. And so that's why roots occur only at the end of words. Otherwise, they might occur at the beginning of words or in the middle of words. Now, in fact, there are a couple of languages where roots occur by themselves without any added material, without reduplications, without dummy syllables, without a knot in compounds. These two languages, of these two languages, one is in Taiwan, it's Boonun, and one is in the Philippines, it's Cebuana. And the form under which these roots occur in these two languages is pretty much the same. You have the CVC structure of the root emerges as CVVC with a global stop between the two Vs. And that looks a lot like type A, Chinese type A. At least what we reconstruct is ancestral to Chinese type A, a sign of Tibetan letter. For instance, in Cebuano, I've cited roots in Cebuano, the equivalent, I mean, the monosyllabic is, well, John Wolf uses a term for these forms, CV, global stop, VC. He calls them stretched. They are stretched monosyllabic roots. The vowel has been stretched, has been germinated, and the global stop has been inserted in the middle. Now, you can, let's move into page five, and table three at the top of page five. Here you find examples from Boonun. For instance, a word, a root, a monosyllabic root, T-A-Q, Earth, which occurs with a non-stretched form at the end of dicinibals, like you see in the second column, like pro-Western Middle Polynesian, buk tak, amis dit tak, supot tak, supot tak, et cetera. But in Boonun, the root occurs all by itself, in stretched form, taak, in the global stop is actually the option. The other examples are like that. You also have evidence of the same alternation in personal pronouns. Free personal pronouns are stretched, and bound personal pronouns are not stretched. Now, let's move to section five by Moray City in Austro-Agiatic. In fact, the exactly, pretty much exactly, the same system is found in Austro-Agiatic, as was observed by Norman Zeid in 2002. Zeid calls it the bimorayic constraint. Zeid was the first to point out the role of bimorayicity in this kind of behavior. According to Zeid in Anderson, exactly the same system can be reconstructed into proto-Austro-Agiatic. And table four gives you examples from four Munda languages, Gutob, Joan, Gorum, and Sora. The way these languages deal with roots, with syllabic roots, with CVC roots is that Gutob reduplicates the first CV. Joan puts a dummy syllable, a dummy E, in front of it. Gorum geminates the vowel, puts a vowel stop in the middle, and Sora does the same as Gorum, but loses the first vowel, just like stage two in figure one. So as they point out, what these languages do is that they have a constraint against monomorayic free-forms. They don't want monomorayic free-forms. They want a free-form to have at least two moray. And there are various ways in which they can achieve that result. Now, our proposal is that the old Chinese type A type B distinction, and in fact, even the prolocynote-betton type A type B distinction as we reconstructed, goes back to a language earlier than prolocynote-betton. That had the same constraint. That had a bimorayic constraint. How do you go from a language that has the bimorayic constraint to all Chinese? Well, you do that by making the dissyllables monosyllabic, and they become monosyllabic by losing their first consonant. So you are left with monosyllabals that have no stretching. And then for the type A words, the stretched forms, they evolve as stretched forms. And there has to be one change, which is you need to change the global stop between the two vowels into a pharyngeal consonant. And I have no explanation for that. It just has to be stipulated. Possibly one could speculate that the languages to the north of Chinese, whichever an ATR versus RTR contrast played a role there, the type A type B distinction being reinterpreted in terms of ATR, RTR. Now, this model implies one thing. It implies that Starleston's proposal there is a correlation between type A type B in Chinese and vowel length in Cukicin is correct. And we have attempted to test that proposal, that hypothesis, statistically. First, we need to set up a null hypothesis that we want to beat. The null hypothesis here is that Starleston is wrong. There is nothing special. There is no particular correlation. There does not exist any positive correlation between Cukicin long vowel and all Chinese type A on the one hand, and no particular correlation between Cukicin short vowel and type B on the other. And like Starleston, we do not compare all Chinese type word type with Lushai, but with Cukicin, the entire group. And for this, we use Kenneth van Beek's 2009 reconstruction of Puddo Cukicin. There are various benefits in that. First, the number of forms is more manageable. And second, single language irregularities tend to be ironed out. So in order to do that, we scan the proto-Cukicin material for Chinese comments. Of course, with some knowledge of a phonological evolution, both into Chinese and into Cukicin. And we had to exclude certain sets of words. For instance, we didn't want verbs, because verbal morphology in Cukicin often affects vowel length in ways that we do not understand. So we left verbs out and worked only from nominals, of which van Beek gives a list. In addition, we excluded open vowel syllables, because there is no length contrast in those in van Beek's reconstruction. So we left those out. We also left out other types of forms and I leave you to look at the handout to find out. Of course, we also excluded words that have both Taipei and Tai-Pi in China. Tai-Pi, Tai-Pi, variants in Chinese, there are many of those, like Roo and Na, and many others. And the same in Cukicin, excluded probable loan words like silver and comparisons requiring large semantic shifts, like the one I give you at the bottom of page 7. In total, we retained 43 comparisons, which are listed in the appendix. And from these comparisons, we built a table, Table 5. Now, from the comparisons we have, there are four possibilities. They can fall into four situations. Chinese type A, Cukicin short, Chinese type A, Cukicin long, Chinese type B, Cukicin short, Chinese type B, Cukicin long. And these four situations are the four cells in Table 5. Now, it is possible, using statistics, to work out the significance of the deviation from the null hypothesis. And a measure of the significance of that deviation is the p-value. With this type of data, it is possible to compute the p-value out of the four figures in Table 5, using Fisher's exact test. And we set the significance level as 0.05, which is the usual significance level in scientific works. Anything below 0.05 is deemed significant. And of course, more stringent significance level of 0.01 would be even more stringent, but we selected the 0.05 level. 0.05, the p-value of 0.05 means that the probability of obtaining the numbers under the null hypothesis is 5%. So anything below 5% is deemed significant. In this case, the p-value is 0.032, et cetera, which means that the probability of obtaining these results under the null hypothesis is about 3.2%, which is low. So at first sight, on present evidence, there probably exists a positive correlation between Cukicin long short and Chinese type A type B. As the established team proposed. Of course, we are well aware that other scholars could have found other coordinates. And then, if that is the case, then it may be necessary in future to modify our list and to recalculate the p-value. However, notice that the predominance of the protocol Cukicin short for the Chinese type B category is very strong and that probably is going to stay. Now, notice that out of the 43 forms in table 5, there are 11 that do not match expectations. I call these mismatches. Now, what is the explanation for these mismatches? The preferred explanation, in our opinion, is that we are not dealing with exact coordinates. We are dealing with reflexes of two worlds which contain the same root. One of them being monosyllabic, originally monosyllabic, monosyllabic, and the other being originally dysyllabic. So that would be the explanation for these mismatches. And of course, the same explanation goes for the variance in both in Cukicin and in Chinese. For instance, ru in Chinese, ru enter noop, would go back to a dysillabal noop, while na would go back to a monosyllabal noop at a pre-Synotibetan level. So that's about it, and I thank you for your attention.