Tangut being a West Rgyalrongic language and the uvularization hypothesis

~ vampyricon

This conclusion is reached entirely based on one paper and the title of another: Uvulars and uvularization in Tangut phonology by Gong Xun (2020) and Tangut as a West Gyalrongic language by Lai Yunfan et al. (2020), which Gong Xun is coäuthor of.


Introduction

Tangut is a Trans-Himalayan (i.e. Sino-Tibetan) language spoken by the people of the Western Xia Dynasty. They have rhyme books/tables, and we have transcriptions of their language into other languages, and transcriptions of other languages into their language. I’m not familiar with the history of the scholarship of Tangut, but what we ended up with is a language with two tones, three “grades”, nine rhyme groups, and an unholy number of vowel qualities, at least according to the traditional reconstruction by Gong Hwang-Cherng. In the interest of clarity, I will be referring to Gong Hwang-Cherng as GHC and Gong Xun as Gong Xun.

GHC reconstructs the three grades of Tangut as having no medial (Grade I), having an */-i-/ medial (Grade II), and having a */-j-/ medial (Grade III). Gong Xun (2020) refutes this by citing transcriptional evidence to and from various languages as well as cognates in other Rgyalrongic languages. He instead proposes the uvularization hypothesis based on Rgyalrongic cognates, in which Grade I has a uvularized vowel, Grade II has a uvularized vowel and a */-ʕ-/ medial, and Grade III simply has a plain vowel. As a result of this, velar initials in Grades I and II are backed to uvulars.

The uvularization hypothesis: Successes and failures

I would like to separate Gong Xun’s uvularization hypothesis (U) into two claims: (U.1) Grade I and II “velars” are actually uvulars while Grade III has true velars, and (U.2) Grade I and II vowels are uvularized while Grade III has plain vowels. I believe (U.1) is true, but I believe (U.2) is false. This belief arose from reading Gong Xun (2020) again after reading Lai et al. (2020) and taking note of the comparative Rgyalrongic data.

In this blog post, I am taking the conclusion by Lai et al. (2020) for granted: I will be assuming that Tangut is a West Rgyalrongic language. Given this, I will show that Tangut Grade I vowels are probably not uvularized.

Gong Xun (2020) reached (U.1) using tables 5 to 7 as evidence. Grade I and II syllables with “velar” initials have uvular cognates, while Grade III has velar cognates. Given that Tangut is a West Rgyalrongic language, the “exceptional” correspondences in tables 8 and 9 are unproblematic. The cognates for which there is Stau and Khroskyabs (both West Rgyalrongic) data available are reproduced below. Japhug, an East Rgyalrongic language, is included because that is the original comparison language. Tangut reconstructions are converted from GHC’s to Gong Xun’s 2020 reconstruction. Tangut characters omitted until I install a Tangut font on my computer.

Tangut (W)Japhug (E)Stau (W)Khroskyabs (W)
𗥦 */ʁuʶ¹/ ‘head’/tɯ-ku/ ‘head’/ʁə//ʁû/
𗝎 */qʰuʶ¹/ ‘bowl’/khɯtsa/ ‘bowl’/qhəzə//qhû/
𘔯 */kʰuu¹/ ‘greet, pick up’/-qru/ ‘greet, pick up’/-ɴqhrə/
𗿈 */kʰij²/ ‘pigeon’/qro/ ‘pigeon’/sqhré/
𘊗 */kiwr¹/ ‘ant’/qro/ ‘ant’/skhru//skhrɑ̂ɣ/
Table 1: “Exceptional” correspondences in Tables 8 and 9 of Gong Xun (2020). Boxes for which the data is not available are left blank. One can see that these correspondences are not actually exceptional when compared with the other West Rgyalrongic languages. (The advances in Gong Xun (2022) have been omitted for clarity.)

From Table 1, it can be seen that the cited Tangut exceptions actually correspond to West Rgyalrongic languages quite well, which further bolsters the reconstruction of uvulars.

However, (U.2) is much less successful. Gong Xun proposed that (U.2) arose due to the compression of syllable-peripheral features such as uvulars onto the vowel, and cites Rgyalrongic uvulars in word-initial or word-final position as evidence of uvularization. The evidence is presented in Table 2:

Tangut (W)Japhug (E)Stau (W)Khroskyabs (W)
1. 𗈋 */tsuʶr¹/ ‘winter’/qartsɯ/ ‘winter’/rtsô/
2. 𗰝 */pʕịʶ¹/ ‘frog’/qaɕpa/ ‘frog’/spəɲcher/
3. 𗼛 */tśʕiʶr/ ‘right’/χcha/ ‘right’ †/rchila/
4. 𗗣 */lwəʶ²/ ‘buy’/-χtɯ/ ‘buy’ †/rə//jdə̂/
5. 𗾔 */biʶ²/ ‘sun’/ʁmbɣi/ ‘sun’/ɣbə/
6. 𗖳 */laʶ¹/ ‘weave’/-taʁ/ ‘weave’/dɑ̂ɣvi/
7. 𗠵 */pạʶ¹/ ‘be thirsty’/-ɕpaʁ/ ‘be thirsty’/spar/
8. 𗑉 */meʶj¹/ ‘eye’/tɯ-mɲaʁ/ ‘eye’/mu/
9. 𗭼 */swiʶw¹/ ‘clear, bright’/-fsoʁ/ ‘bright’/fsu/
Table 2: Evidence cited by Gong Xun (2020) for reconstructing uvularization in Tangut vowels. Boxes for which the data is not available are left blank. One can see that there are no uvular segments in the West Rgyalrongic cognates.
† The first phone is a voiceless uvular fricative. It seems the font is causing some issues.

We can see that there are no uvulars to be seen in the West Rgyalrongic cognates, so I doubt that the hypothesis (U.2) is correct, unless 1. the uvular segments in East Rgyalrongic can be reconstructed to Proto-Rgyalrongic, 2. West Rgyalrongic splits into two branches early on, with Tangut on one branch and Khroskyabs-Stau on the other, and 3. Tangut preserves the ancestral uvular segments, only to have them transfer onto the vowel. I also cannot see any unifying feature that might explain what Grade I rhymes in syllables with non-dorsal initials (velar and uvular) actually are, at least, none that pop out to me.

Table 2 may provide a hint as to the nature of Grade II rhymes in syllables with non-dorsal syllables, however. I would emphasize that it is a hint, and not the answer, because there are only two items in Grade II cited. However, the Stau cognates of both exhibit palatal segments, in particular, /ch/. If this holds across more cognates, perhaps the feature that determines Grade II syllables with non-dorsal initials is a palatal medial. Of course, this would raise the question of what the unifying feature of Grade II is, or if there even is one. On the other hand, Table 6 of Gong Xun (2020) shows that Stau cognates to Tangut all show either a /r/, a /z/, or a /s/ adjacent to the uvular segment, so perhaps something around the alveolar region is going on.

Conclusion

Gong Xun’s Tangut uvularization hypothesis comes in two parts: 1. dorsal initials in Grades I and II are uvular while in Grade III they are velar, and 2. the feature that distinguishes Grades I and II from Grade III is the uvularization of the vowel. This uvularization arises from the features at the periphery of the word transferring onto the vowel, referred to as “compression”. However, when one examines West Rgyalrongic cognates of Tangut, which itself is a West Rgyalrongic language, one fails to find support for part 2 of the uvularization hypothesis, as the cognates to Tangut Grade I and II syllables without dorsal initials do not have uvular segments.

Update

I have now installed a Tangut font and have added Tangut characters to the tables. Fixed the citations.

Bibliography

Leave a comment