 Good morning and welcome to my presentation. First of all, I would like to thank my mentor, Nathan, for inviting me to talk to you here today, and I hope that I'll be able to present you with something that you would find interesting to listen to. I would also, from the outset, like to invite your comments and suggestions. I have no handouts, but I have two papers, draft papers, which I'd like to share with you if you are, after this talk, interested to know more. I'm basically a field linguist. I've described a small and endangered Tibetan-Burman language spoken in the Eastern Himalayan region. I've written a grammar of that language, and then during my field research, I also collected data on seven presumably related linguistic varieties, and right now I'm using these data to reconstruct the purported proto-language, which I've called proto-Western Kobwa, and today I would like to share with you a few ideas on the onsets, looking from the perspective of two of these varieties. I'll start with a short introduction, because probably the Kobwa languages are, you are not very familiar with them, if you've ever heard of them at all. I will give a short introduction to my data and the methodology that I followed, which all of you are very much familiar with, so I'll not discuss that in detail. I'll quickly show a few trivial and regular correspondences. Again, not very surprising, not very interesting, so I'll just go through them very quickly. And then I would like to focus on some of the more extraordinary, not really maybe in a typological perspective, but at least in this case, what caught my attention, which are the pellet-laced onsets, the rothic onset clusters, juvenile onsets, voiceless onsets. And finally, I would like to show you a few noteworthy correspondences with other Tibetan-German languages. Kobwa languages are a small cluster of supposedly related languages spoken in Western Arunachal Pradesh, which is the area which is bordered by Bhutan in the West, Myanmar in the East, then the Tibetan Plateau and the Himalayas in the North, and the Brahmaputron Plains of Assam in the South. Based on research materials that became available basically only in the last two decades of the previous century. It was Jenshin Jackson-Sun who first came up with the hypothesis that these languages, instead of being just some kind of a residue of languages, Tibetan-German languages actually might be related to each other. Since then, his view has been both adopted as well as challenged. However, till now, there has not really been a lot of or any linguistic evidence which either supports or discredits his view. Among these Kobwa languages, a certain subgroup can be discerned, the Western Kobwa languages, which are mentioned here, basically four languages with a total of eight individual varieties spoken by less than 7,000 people in the Western part of Arunachal, with as close as genetic relatives Puroik and Bugun. All these languages must be considered endangered, not just because of the low-speaker population, but also because of the rapid socio-economic developments in this area, which are really very rapid. This map shows all the hitherto identified Kobwa languages, the Puroik area, which is in the central and western part of the state, huge geographical area. They are speaking this language interspersed with Nishibangru and Miji speakers. Here, we find the closely related mutually intelligible Bugun varieties. Here in the western part of the area, we find the Western Kobwa languages. Kisfi and Duhumbi spoken here. Then the Sautang varieties spoken in this small area here and the Sherupen variety spoken in the Tengah river valley. As you can see, this is a relatively fragmented area. There are speakers of other more dominant both linguistically as well as socio-culturally more dominant speakers. In the neighborhood, the Tangla speakers, for example, who separate the Kisfi and Duhumbi from the other western Kobwa speakers. There has been a lot of language contact, both through contact languages as well as to inward migration and intermarriage, not just from Tangla also, from Central Bolish Brokpa, from the Tibetan side, Tuchangacha speakers from Bhutan, Kengah speakers from Bhutan and in the eastern part the Miji and the Russo Aka. So a lot of contact language influence, which is important to be aware of. My former colleague Ismail Lieber here and myself, we did a lexico statistical study, a Kognut count analysis with sensitivity analysis and here you can see the results. We used 100 so-called stable roots and around 20% of these stable roots are shared among all the Kobwa varieties. A percentage which goes up to around 40% between two individual Kobwa varieties. So that's not a lot but in an area like this with so much of language contact, considering stable roots which are not very prone to borrowing that might be indicative. We can also see that the western Kobwa languages are the square in the left upper corner, Bugun languages in the middle and the Poloik varieties which show most internal variation in the right lower corner. So there appears to be a rather clear subgrouping. My research is based on original field data of all the eight known western Kobwa varieties. I started out with a word list of more than 550 lexemes which were from the most commonly used word lists but also regionally, socioculturally, environmentally specific words which I collected based on my own experience in the area. A minimum of two speakers from each variety and I also collected additional lexemes based on sentences etc. to find more about the morphology of these varieties. Finally I went back to elicit some missing lexemes like the gaps in the cognate sets and also some interesting forms on basis of particular onsets or particular rhymes which I told that might be interesting to have some more information about. Methodology you will be very familiar with. One important thing to mention is perhaps that I started out with a comparison between Duhumbi and Koitam and the reason for that is that Duhumbi is rather conservative as far as rhymes are concerned. It has innovated a bit more in the onsets and it's also been on a more contact language influence but still the rhyme patterns are very interesting and Koitam is rather representative of the Sautang and Shendepen varieties. Here are some trivial correspondences nothing out of the ordinary here in which all the Western Cogba varieties have the same reflex from a reconstructed protoform. Regular correspondences where they don't all have the same reflex but still these reflexes are not very surprising, not very special and easily derivable from an onset. So rather than that I would like to focus a bit more on some special onsets. First of all there are a number of cognate sets where Duhumbi and Kispi have a voiceless palatal fricative onset which corresponds regularly with aspirated palatal and alveolar africates in the Sautang and Shendepen varieties. So my proposal here is that this derives from palatalized plosive onsets both in the voiceless both the voiced and the voiceless counterpart there in which it is the Sautang and Shendepen varieties which actually show a distinction between an aspirated reflex and a voiceless reflex. There is another correspondence set where a Duhumbi simple bilabial plosive onset corresponds to voiced palatal africate onset in Koitam. However, this is only when followed by an open high vowel and this is actually a homophonous set of a homophonous correspondence four cognate sets in which we find this correspondence which I've attributed to the open high vowel and as you can see in the second set of examples the influence of an open high vowel or open high vowel rhyme on an onset can be rather distinctive. So whereas normally we have a correspondence of a bilabial plosive in all varieties there are a few cognate sets where we see a bilabial plosive in Duhumbi and Kisvi correspond to the alveolar africative, voiced alveolar africative, the sibilant onset. So here we can see this distinct influence of high vowels and high vowel rhymes. You may notice that there is a kind of a noticeable correspondence between western ko-boa bilabial plosive onsets and nasal onsets in the Tibetan language. This is something that I will come to a little bit later. Then the second is the robotic onset clusters. There are certain correspondences that are irregular in the sense that the correspondence itself is regular. The cognate sets are regular but there is a different kind of reflex in some of the varieties. For example, here you can see that in the first group that the ko-itam reflex is an africate whereas we would have expected a simple velar onset. So in this case I have reconstructed the robotic onset clusters both with velar stops, with bilabial stops and with a voiced alveolar africative onset. So interesting is perhaps the phonological change that affected these onsets. These are regularly simplified in the Humbi-Kispi and the Sharedupen varieties to a velar onset but they became palatal africates in ko-itam, raung and jirigau and there are very peculiar retroflex africates at least in an aerial perspective for the western Kogwa languages. It's more common in Rushi's languages of the retroflex africates. I have only one example from the voiced onset and there are several examples from aspirated velar onsets. The reflexes are a little bit different in the case of a voiceless, an aspirated velar onset. There is a glottal africative onset in jirigau with a subsequent contraction of the root and the prefix. So in the first case of the wild boar se-hen becomes sen. In the second case, barking deer se-hi becomes si. But we see that the glottal africative onset has been preserved in the case of bone which might be attributed to the vocal prefix instead of the prefix's sha, the sha prefix, the meter animal prefix. So this is the only case in which there are a little bit unexpected reflexes we would have expected, for example, unaspirated, unvoiced reflexes also in Duhumbi, but as you can see in the case of barking deer that correspondence holds. But in the case of wild boar and bone, there is aspiration, which may be attributed to a prefix. Clusters of a bilabial, plosive, and rotic medial are simplified to bilabial, plosive, onsets in Kisbien, Duhumbi, but have been preserved in the sartang and shedepen varieties, although the rotic became a lateral medial in most sartang and shedepen varieties except for Rahum, which has, in most cases, well, in all cases, preserved the rotic medial. The reason for not reconstructing a bilabial, plosive, and lateral onset cluster here is because there is no other evidence for lateral onset clusters. So veeler and lateral onset clusters, I have no evidence that suggests this, but as you saw in the previous slides, there is evidence for a veeler and rotic medial onset. And I've earlier shown that bilabial onsets followed by a high vowel or high vowel rhyme have divergent onset reflexes. The same is the case with these rotic onset clusters. So the reflexes here are different from what we would expect. So again, it is this high vowel or high vowel rhyme, which seems to have some kind of deviating influence on the onset. These correspondences are the main source of the Duhumbi onset cluster, pshe, which indeed itself occurs only when followed by a high vowel or a high vowel rhyme. In the case of swell, fill, full, and four, the onset reflexes actually, in my opinion, may not really show for a onset cluster, but perhaps rather a prefix or a pre-initial rather than the onset cluster. But I'm sure there are people here who have their own interesting ideas on that, so maybe we'll keep that for discussion later. Now I would like to pay some attention to a rather modally mixed bag of correspondences that I was kind of left with after setting up all the other correspondence patterns. We see that the semantics of these sets are very clear. There is not much case where you have to look for semantic cognate forms, which are very different. And even the rhyme reflexes are rather predictable. They are regular. So the onsets are the only ones that are a little bit strange. And based on the comparative evidence, well, first after internal reconstruction, like after excluding other possible reasons, for example, other possible simple onsets, relativized onsets, rotic onset clusters, or possible explanatory phonotactic conditions, it is the comparative evidence where I found some information that this might actually derive from uvular onsets. In the Western Kogwa varieties themselves, uvular onsets are not attested, except in Duhumbi where it occurs an allophone of a aspirated velar onset. For the rest, they are not a distinctive phoneme in any of the Western Kogwa varieties. So like I said, both the internal evidence and the form of onset correspondences that do not fit in with any of the other correspondences and comparative evidence from, in this case, quite often old Chinese, or basically always old Chinese, suggests that in these cases, rotic onset clusters of a uvular plosive can be reconstructed. And in this case, it's the unexpected rhyme reflexes, in the case of soybean and egg, which actually necessitate a labialized onset. And in combination with the divergent onset reflexes, I propose that these derive from a labialized uvular stop, which in the case of koi tam suk has actually been contracted. So there is, again, the prefix sha hua, which becomes se yuk in koi tam and is then contracted to suk. For example, in Jiri Gaon, we still find the original form se yuk. And in the case of hand arm, it is probably the complete absence of a prefix that resulted of the loss of the onset in the satan and shared the pen varieties, where they have now retained only a uvular vocal onset instead of the expected palatal glide. So it would have been yuk if there had been a prefix, but because there was no prefix, the palatal glide was lost and it became yuk. There are a few cognate sets where a Duhumbi and Kispe labial approximate onset corresponds to palatal approximate initials in satan and shared the pen. I propose that this derives from voiced uvular onsets, which, again, is kind of supported by the comparative evidence from Chinese, old Chinese. There is another correspondence where vocal onsets in the western cobra varieties correspond to voiced uvular onsets in old Chinese, which I will show you in a few slides. There are several cognate sets between western cobra varieties that cannot be explained. And here I would like to reconstruct voiceless onsets in which the characteristic reflex is actually a glottal fricative and at least one of the varieties. Why, in the case of voiceless nasal onsets, the glottal fricative is in Duhumbi, whereas the nasal is in the satan and shared the pen varieties. But it's exactly the opposite when we look at the approximate onsets, in this case the palatal and the labial approximate onset. That is something that's still a bit of a mystery to me. Here are some of the examples. I think considering the time, I will not go much into these. In the following slides, I'd like to present some additional evidence of relatively unique correspondences between western cobra and other Tibetan-Burman languages. Some of these correspondences were already identified before, and I would like to present some additional evidence for these. Characteristic for the cobra languages is the correspondence between plosive onsets with nasal onsets in other Tibetan-Burman varieties. I saw already in your handouts that that also is a correspondence between Basque and Proto, or considered to be a correspondence between Basque and Proto-Indo-European, so it might not be that typologically that strange. But for the Tibetan-Burman languages, except for a few Bisou varieties, this is probably quite exceptional. Some examples of these correspondences are provided here with the western cobra reconstructions, having a plosive onset, but the comparative evidence from other Tibetan-Burman languages showing a nasal onset. This correspondence is also found in the Vila point of articulation, as these examples show. Here a western cobra-Vila plosive onset corresponds to other Tibetan-Burman-Vila nasal onsets. And there's even one case at the dental point of articulation, although the evidence is only for Duhumbi and Kispi, because the Sartang and Sherdupan varieties have a nasal onset as the other Tibetan-Burman languages. I have noticed how somehow correspondences of Proto-Western Cobra stop onsets often tend to correspond to all Chinese nasal onsets in type A syllables. Some of these examples I already showed before On the other hand, Proto-Western Cobra nasal onsets often correspond to all Chinese nasal onsets in type B syllables. So this is perhaps a kind of a noteworthy correspondence. However, as always, where you find regular correspondence patterns, you have your exceptions. Where either Proto-Western Cobra nasal onsets correspond to all Chinese nasal onsets in syllables of type A, where we can see that these syllables pelletalize, the onsets pelletalize in western cobra, or we have Proto-Western Cobra stop onsets that correspond to all Chinese nasal onsets in syllables of type B. Now, because I'm still a little bit unfamiliar with the distinction between all Chinese type A and type B syllables, and the various theories that surround this, I cannot yet draw any conclusions regarding this, but these are just my observations, and I'm happy to hear people's ideas and suggestions. Another observation is that western cobra vocal onsets often correspond to a sibilant onset in other Tibetan Berman languages. Some of the cognate sets that show these correspondents are presented here. A major outstanding issue is shown in the last two examples where a voiceless LVL or fricative onset in other Tibetan Berman languages just regularly corresponds with a sibilant onset, or in some cases a glottal fricative onset in the western cobra varieties. So what are the conditions that result in de-bucalization or in preservation of the original onset? This is again something that I will have to look into more closely. There is another set of correspondences where western cobra vocal onsets correspond to reconstruct all Chinese voiced uvular stop. In these cases, although it's tempting to reconstruct a voiced uvular stop for proto-western cobra, I think it is prudent to just reconstruct the most straightforward onset, which in these cases would be the vocal onset. Maybe later evidence will actually show that this derives from all uvular stop onsets. Then there is a regular correspondence in the western cobra languages between voiceless dental nasal onsets initials. Two examples of this are given here. In some cognate sets, we can observe how the voiceless dental onset in western cobra corresponds to a voiceless lateral or voiceless or voiceless lateral onset in type A syllables in all Chinese. So again, just an observation of quite a regular pattern, at least in those cognate sets where I could find comparative evidence from all Chinese. Although you may not really agree with all my semantic correspondences here. A rather curious correspondence that is nonetheless very regular is the correspondence between a Duhumbi and Kispy dental plosive onset and erotic onset in all other varieties. I propose that this derives from an onset cluster of voiced uvular fricative and erotic medial. As you already saw, erotic medial onset clusters are not uncommon, at least in the reconstruction. It's also noted how this western cobra correspondent quite often tends to correspond with dental plosive onsets in all Chinese. So again, an observation of a possible comparison with all Chinese and other Duhumbi languages. Okay, wow. Sorry, did you send that off? You didn't have to use the... Okay, thank you very much for your attention. If there are any questions, suggestions, ideas, then I would be happy to discuss them since there is more than enough time anyway.