12.12.22.23:59: THE HUNDRED NOTATIONS OF HANGUK: COMBINOTATION

I conclude my series on Korean musical notation with a glimpse at 合字譜 hapchabo 'combined character notation' or 'combinotation'.

Hapchabo is based on Chinese 減字譜 jianzipu 'reduced character notation' fron the late Tang Dynasty, shortly before the invention of the Khitan small script circa 925 AD. Jianzipu contains

only the important elements of the characters (like [qin] string number, plucking technique, hui number [for each of the thirteen dots on a qin] and which finger to stop the string) and combined them into one character notation. This meant that instead of having two lines of written text to describe a few notes, a single character could represent one note, or sometimes as many as nine.

Here is an 1864 sample of jianzipu. The symbols superficially resemble Khitan small script blocks, though they include partially enclosing components absent from Khitan: e.g., the second, fourth, and sixth of the symbols in the second large column from the right are

⿺ 乚 (short for 挑 indicating the right index finger) with 五 'five' on top

⿹ 勹 (short for 勾 indicating the right middle finger) with 七 'seven' below

⿰ 夕+十 'ten' atop ⿹ 勹 with 四 'four' below

If these components existed in the Khitan small script, they would be combined as

乚五

勹七

夕十
勹四

without moving the numbers into the open spaces of 乚 and 勹.

The Hangul letter ㄱ (but not ㄴ!) allows some vowel letters in its open space: e.g., 고 (not ㄱㅗ) and 교 (not ㄱㅛ).

This type of combination in jianzipu and Hangul is inherited from sinography: cf. characters like 旭 and 旬. It makes jianzipu block generation more difficult than Khitan small script block generation. (I wonder if Candy Yiu has made any progress with jianzipu rendering since 2005.)

From the few examples I have seen, hapchabo has a somewhat different structure. There are three core elements per block read from top to bottom, right to left (unlike Khitan small script or Hangul which are read from left to right, top to bottom) with lines or dots added to either side: e.g., this block

which Kim Jin-Ah (2010: 206) explained as follows (italics added and capitalization and romanization changed to fit the style used on this blog):

In this symbol, the letter tae (大) is the abridged letter for taehyŏn (大絃 ['big string']), six (六) indicates the 6th fret, kiyŏk (ㄱ) is the abridged letter for the [left] thumb, and the left stroke (ㅣ) is fingering code instructing the musician to beat from munhyŏn (first string from the player) to yuhyŏn (second string) or taehyŏn (third string). In other words, it instructs the player to press taehyŏn at the 6th fret with the left thumb, and with taehyŏn being pressed, beat with the right hand from munhyŏn to yuhyŏn or to taehyŏn.

Although Kim referred to the thumb symbol ㄱ as "kiyŏk", the name of the Hangul letter ㄱ <k>, I am not sure that ㄱ was originally meant to be a Hangul letter. In examples of hapchabo I have seen, it seems to correspond to ㇆ (but with a bent line leading to the hook) which is similar to the jianzipu symbol 勹 (for the right middle finger, not the left thumb!). So perhaps I was wrong to regard hapchabo as a hanja (Chinese character)-Hangul hybrid, at least in its early stages.

Apparently modern Korean musicologists treat hapchabo symbols resembling Hangul letters as if they were those letters: e.g., in examples of hapchabo I have seen, the left middle finger symbol now called "niŭn", the name of the Hangul letter ㄴ, seems to correspond to レ which resembles the jianzipu symbol 乚 (for the right index finger, not the left middle finger!). Could hapchabo be said to be Hangulized, and if so, when did that happen? Does modern hapchabo have Hangul letters instead of the older non-Hangul symbols ㇆, レ, etc.?

12.23.1:50: If what are now considered Hangul letters in hapchabo were originally Hangul letters, I would expect them to

- form a set of nothing but Hangul letters

- be in some variant of Hangul letter order: ㄱ, ㄴ, ㄷ, ㄹ, ㅁ, ㅂ, ㅅ ...

- or be the first letters of words for 'thumb', etc.

But according to Lee Jin-weon (2010: 154) the hapchabo left finger symbols are

I have added the non-Hangul symbols for the thumb and middle fingers.

There are other less common finger names, but among them 巨指 kŏji 'giant finger' is the only one whose initial letter matches a Hangul hapchabo symbol.

Contrary to what I just wrote above, I now think the five symbols could be derived from Chinese finger names:

One flaw with this theory is that 中 does not contain the strokes ㄴ or レ; its four strokes are ㅣㄱㅡㅣ. But ㅣ could be confused with the vertical line for the fingering code, ㄱ could be confused with ㇆ for the left thumb, and ㇐ looks like 'one' for the first fret. So ㄴ~レ may have been chosen as an abbreviation that resembles 中 while not risking confusion with any other symbols.

12.12.21.23:59: THE HUNDRED NOTATIONS OF HANGUK: CARNOTATION

In the previous installment, I wrote about the Korean use of sinographic elements for new purposes in yulmyŏng 'pitch names'. Although some yulmyŏng characters like 㶂 are uniquely Korean, all of their components are Chinese.

These next two installments will deal with two systems of Korean musical notation containing hanja (Chinese character)-Hangul hybrids.

肉譜 Yukpo, literally 'meat notation' (carnotation with haplology?'; 'mnemonic notation' in Kim Young-woon 2010: 23), consists of names written with five types of symbols (though not all five were used simultaneously):

1. hanja proper (i.e., Chinese characters imported from China without modification)

- with Sino-Korean readings: e.g., 至 chi

- with native Korean readings: e.g., 浮 ttŭl* (ttŭ- is the native verb 'to float'; -l is a verb ending used in native Korean tags for hanja: 浮 is ttŭl pu 'float pu'; pu is its Sino-Korean reading)

2. abbreviations of the above hanja: e.g., 至 > 𠫔 chi

3. 國字 kukcha (made-in-Korea hanja)**: e.g.,

乷 sal***, a combination of 沙 sa atop 乙, a common phonetic element for -l in kukcha (which also happens to resemble Hangul ㄹ -l, though the use of 乙 to represent a liquid coda long predates Hangul on the Korean peninsula)

㐍 ttŭl****, which I think is a combination of 爫, an abbreviation of 浮 ttŭl (see above) and 乙 (see above)

4. hanja-Hangul hybrids for syllables that are not readings of sinographs: e.g.,㪳 tung from the sinograph 斗 tu plus the Hangul letter ㅇ ng

5. Hangul syllables as substitutes for any of the above: e.g., 지 chi, 뜰 ttŭl, 둥 tung

I last wrote about hanja-Hangul hybrids three years ago. At that time I didn't know that they were also used in musical notation. I assume they originated as transcription characters that were later recycled in yukpo.

*12.22.1:09: 'Float' was [ptɯrʔ] in Middle Korean. I do not know whether the yukpo names predate or postdate the shift of pt- to tt-. Until the 20th century, tt- was spelled as ㅼ <st>, so one could call this yukpo name stŭl.

**12.22.1:15: Hanja-Hangul hybrids (see above) are also kukcha, but I distinguish them from kukcha like 乷 which consist solely of existing sinographic components (沙, 乙) in new combinations. Yulmyŏng characters like 㶂 may be considered the latter type of kukcha, though one might argue that they are not linguistic symbols.

***12.22.00:59: Oddly 乷 is short for 乷冷 ssaraeng in which 乷 is read as an open syllable ssa. Was the original name *沙冷 ssarayng (ssaraeng in modern pronunciation), abbreviated as *ssar and then pseudo-Sino-Koreanized as sar (later sal)? (*ssar is not a possible Sino-Korean reading but sar was and sal is. I don't know when final -r shifted to -l in Korean. If the abbreviation occurred after the shift, ssaraeng could have been shortened to sal.)

*沙冷 ssarayng  > ssaraeng : 乷 *ssar > sar > sal

Then the graph 乷 for sar ~ sal was used as the first graph of ssaraeng, making the graphic relationship between the short and long names more obvious:

*乷冷 ssarayng  > ssaraeng : 乷 *ssar > sar > sal

****12.22.1:24: zdic.net lists the Mandarin reading zhǎo for 㐍, but the Unihan database only lists a Korean source for the character. (The Kangxi reference in the Unihan database ends in 1, indicating a theoretical location for the character in that Chinese dictionary.)

I doubt 㐍 was ever used in Chinese. I cannot find it in the Taiwanese Dictionary of Chinese Character Variants (which I admit is incomplete: e.g., it lacks a four-stroke variant of 人 resembling the Tangut character for 'person'). The closest match was ⿺ (not ⿱!)  爪+乙, a variant of 糾 jiū. I suspect that the reading zhǎo for 㐍 is from 爪 zhǎo.

Next: Combinotation

12.12.20.23:59: THE HUNDRED NOTATIONS OF HANGUK: YULMYŎNG

I have only barely begun to look through Hwang et al.'s Musical Notations of Korea (2010; available online in its entirety for free). It does not cover all of the "more than 89 types" of notation that are known (Kim Young-woon 2010: 32), but I am grateful for the opportunity to begin to explore a world I know nothing about.

Here is my attempt to reconcile and integrate what I've read in chapters 1 and 2 with these two HWP files (authorship unknown; I can't identify the sites that presumably link to them):

The 律名 yulmyŏng 'pitch names' for the twelve basic 中聲 chungsŏng 'middle sounds' are all written with standard Chinese characters: e.g.,

黃 hwang (short for 黃鐘 hwangjong)

However, 淸聲 chhŏngsŏng 'clear sounds' (pitches an octave higher) could be indicated in the early 15th century by adding 淸 'clear' to the bottom right of a yulmyŏng: e.g.,

黃_淸 chhŏng hwangjong 'clear hwangjong'

By the late 15th century, 淸 chhŏng was abbreviated to 氵 which was identical to the sinographic radical 氵 su 'water' and written on the left like that radical: e.g.,

潢 chhŏng hwangjong 'clear hwangjong'

By 1906, pitches two octaves higher were written with two 氵 on the left: e.g.,

㶂 chungchhŏng hwangjong '重淸 heavy clear or '中淸 middle clear* hwangjong'

Conversely, by the 20th century, 倍聲 paesŏng 'double sounds' or 濁聲 thaksŏng 'muddy sounds' (pitches one octave lower) were written with 亻, presumably an abbreviation of 倍 pae* which was identical to the sinographic radical 亻 in 'person' and written on the left like that radical: e.g.,

僙 pae hwangjong '倍 double hwangjong' or thak hwangjong '濁 muddy hwangjong'

Pitches two octaves lower were written with 彳 which might be an abbreviation of 亻亻 ; it is identical to the sinographic radical chhŏk 'step' and written on the left like that radical: e.g.,

㣴 paethak hwangjong '倍濁 double muddy hwangjong' or habae hwangjong '下倍 lower double hwangjong'

This recycling of the shapes of sinographic radicals for new purposes reminds me of the use of those shapes in the Khitan and Jurchen large scripts without any obvious reference to water, people, or steps:

Khitan large script characters

Jurchen large script characters (no 'water' or 'step' lookalikes)

(The second character <dei> looks exactly like Khitan <pa> which is obviously related to Liao Chinese 伐 *faʔ with a dot.)

There cannot be any direct connection between the Korean, Khitan, and Jurchen uses of those shapes since the Korean practice long postdates the extinction of both the Khitan and Jurchen scripts. Nonetheless, I wonder if the creators of the Khitan and Jurchen scripts (and/or their common prototype) also used old components in innovative ways, just as Shong Lue Yang was probably inspired by the appearance of Lao superscript signs to create integral parts of some of his Pahawh Hmong characters: e.g.,

Lao ດ dɔː > ດິ di (change of vowel)

similar-looking second stage PH characters: <ko> > <ko> + ㅡ on top = <koj> (change of tone)

Lao ລ lɔː > ລິ li (change of vowel)

similar-looking second stage PH characters: <nch> > <nch> + ㅗ on top = <nb> (change of point of articulation and deaspiration: [ᶮɟʱ] > [ᵐb])

Yang was not originally consistent in his use of pseudodiacritics in second stage Pahawh Hmong: e.g., ㅡ on top could represent the tones -b, -j, -v, -s, or -g depending on the vowel symbol beneath it. A circle on top could indicate the tones -b, -m, -j, -Ø, or -s depending on the vowel symbol beneath it, and distinguished <nr-> from circleless <v->, <z> from circleless <hl->, etc.

One might expect Chinese radical lookalikes to have similar random functions in the Khitan and Jurchen large scripts, but the creators of those scripts (or their Urschrift) were at least semiliterate in Chinese, whereas Yang was illiterate (though certainly aware of writing and influenced by the scripts around him). Could the creators of siniform scripts have been like beginning students of Chinese, Japanese, or Korean with a knowledge of a small number of standard characters but without knowledge of character structure? They could have then created more characters on their own by

- taking existing characters they didn't know and assigning new functions to them: e.g., 休 *xiu 'rest' as a Jurchen phonogram <li> (cf. the use of the shape of the Roman letter M for <nth> in all stages of Pahawh Hmong).

- combining existing strokes and elements in un-Chinese ways

One problem with this scenario is that it predicts that the inventor of Jurchen knew the sinograph or Khitan large script character 南 'south' - possibly the basis of Jurchen

<fanti> (later <fan>) 'south'

- but not the simple sinograph 人 'person' since the Jurchen character for 'person' was

<niyalma>

with the 亻 'person' radical. How many students of Chinese, Japanese, or Korean can more or less write 南 but can't remember 人?

Similarly, the sinograph for 'south' is also in the Khitan large script

but 'person' is written with a lookalike of a homophonous sinograph 仁 'benevolent'. How many students of Chinese, Japanese, or Korean can write 南 and 仁 but can't remember 人?

*12.21.2:03: I am surprised by this usage of 中淸 'middle clear' because i would guess that it refers to a pitch that is between 中 'middle' and 淸 'clear', not a pitch that is even more clear (i.e., less like 'middle') than 'clear'.

*12.21.3:18: 倍 pae 'double' is a better candidate for abbreviation than 濁 thak 'muddy' because the left side of the latter already signifies 'clear' and the right side is too complex.

12.12.19.23:59: THE HUNDRED NOTATIONS OF HANGUK (PRELUDE)

Thanks to Andrew West who found this passage from Grinstead (1972: 8) that I had remembered incorrectly. It dealt with Hangul, the Korean alphabet, not the scripts of the Tangut, Jurchen, or Khitan:

A civilization so complex as the Korean had, no doubt, more disciplines that could provide a stimulus for the creation of written forms. We would not altogether disagree with King Sejong [the inventor of Hangul] that divination had its part to play. Music and music notation probably had an even more important role, as the script bears a striking resemblence to the pipa (lute) notation. (10)

Note 10 also did not involve the sinoform scripts of Central Asia and Manchuria:

Medieval music notation could also have been influential in forming the Japanese katakana, and the symbols could have been derived from parts of characters before they came to Japan. On the other hand, the notation for the qin (koto) could have influenced the more complex systems of writing, like Vietnamese, in which whole Chinese characters can combine in pairs to form one single written form that can give both sound and meaning.

I think the form (but not sound values or structure) of Hangul letters may have been influenced by the abbreviated Chinese characters of 口訣 kugyŏl: e.g.,

Hangul ㄱ <k> resembles a kugyŏl character for <n> abbreviated from 隱

Hangul ㅈ <c> resembles a kugyŏl character for <myŏ> abbreviated from 彌

But it is hard to be sure because anyone familiar with the sinographic repetoire of strokes would come up with similar shapes without knowing musical notation, kugyŏl, etc. Katakana superficially resemble kugyŏl but developed independently in Japan. Although the man'yougana underlying katakana are of peninsular (specifically Paekche) origin, the simplifications and/or sound values are not: e.g.,

- the left side of Middle Chinese 阿 *ʔa was simplified to 卩 <a> in Korea but to ア <a> in Japan

- the left side of Middle Chinese 加 *kæ 'add' was simplified to 力 in both Korea and Japan, but in Korea it became a phonogram <tŏ> (with a reading taken from the native Korean word for 'more'; i.e., add-itional) whereas it became the katakana <ka> in Japan

Some identical forms have different sources in kugyŏl and katakana: e.g.,

ソ is <hă> in Korean from 爲 'do' (the native Korean root for 'do' was hă-) but <so> in Japanese from 曽

ロ is <ko> in Korean from 古 but <ro> in Japanese from 呂

There are (nearly) identical forms with identical sound values in the two scripts, but they could have arisen by chance: e.g., kugyŏl 刂 <ri> and katakana リ <ri> are both from Middle Chinese 利 *lih, and anyone familiar with Chinese transcriptional practices would probably choose 利, a common sinograph for Indic transcription, for a foreign ri or li and abbreviate

- it as a whole

- its left side

- its right side

So it is not surprising that Korean 俗作 sokchak* 'vulgar-made' musical notation superificially resembles kugyŏl and katakana: e.g., リ, the symbol corresponding to Western A sharp and B, resembles

- kugyŏl 刂 <ri>

- katakana リ <ri>

- the right side of Jurchen characters such as

and the right side of the Khitan large script characters

which I wrote about in "No Kidd-in'"

- the left side of Tangut characters such as

- the Japanese signs for lute tablature sign III-0 (ultimately from 三, not 利)

Sokchak is just one of an estimated hundred types of Korean musical notation. Only thirty were known in 1948, and now "more than 89 types" (Kim Young-woon 2010: 32) are known. What are the others like?

Next: A Small Sampling of Korean Musical Notation

*12.20.1:15: The name is "Sokjak" in Kim Young-woon's (2010:32) romanization.

It also appears as "Sokja" (sokcha in my romanization; presumably 俗字 'vulgar character') in a table on the same page.

Another name for that notation is yakchabo (Yakjabo in Kim Young-woon's romanization; presumably 略字譜 'abbreviated character notation').

12.12.18.23:59: DIE VERLORENE URSCHRIFT DER MANDSCHUREI?

In the mid-90s, I read a passage in Eric Grinstead's (1972) Analysis of the Tangut Script in which he mentioned a possible connection between musical notation and one or more of the new scripts of the post-Tang Sinosphere. (I don't have Grinstead's book on hand at the moment. Is my memory wrong?) Over 15 years later, I was stunned to see Andrew West provide evidence for such a connection:

[T]here are some interesting correspondences between the Jurchen [large script characters for] numbers and [lute] tablature signs [...] The correspondences are far from conclusive, and we have to assume that the assignments of lute tablature signs in extant manuscripts do not accurately reflect their original sequential order. Nevertheless, I think that the similarities are more than just coincidence, and that there must be a common origin for at least some of the lute tablature signs and Jurchen number characters.

Andrew proposed

that lute notation was developed by non-Chinese speakers who used a script based on simplifications and cursive forms of Chinese characters for writing their own language [...] This, now lost, urscript could have been ancestral to both the Khitan Large Script and the Jurchen script, which would explain why Jurchen has many characters that are similar to KLS characters, although Jurchen does not seem to have been directly derived from KLS.

Could this Urschrift have been based on northeastern 'vulgar' variants of Chinese characters? What if some of the early diversity of Chinese writing - which one can sample by looking up the oracle bone, bronze, and seal forms of characters at chineseetymology.org (85 forms of 車 alone on the front page!)- had survived on the periphery to be modified and elevated to national script status by the Khitan and the Jurchen? I think the Khitan and Jurchen scripts need to be examined by a true expert in sinography. Comparing their characters to mainstream standard sinographs can only get us so far.

12.19.00:54: Is it possible that the Khitan large script character 仁 <ku> 'person' is in fact a descendant of one of the more complex early variants of 人 rather than a borrowing of the standard character 仁 to write the Khitan translation equivalent of its Chinese homophone 人? In 2000 I played with writing early forms of sinographs using modern stroke types and the results looked vaguely like the Khitan and Jurchen scripts.

The Khitan small script character 几 <ku> also resembles Chinese 人, though this may be a coincidence.

12.19.3:19: I spent hours looking for this example of a modern obscure variant of 谢 which "possibly goes back centuries". 讠+身+又 has a near-lookalike with 攵 instead of 又 in the Liao dictionary 龍龕手鏡. It just occurred to me that a Liao dictionary might contain at least a few variant sinographs that are ancestral to Khitan and/or Jurchen large script characters.

12.12.17.23:57: BIRDS OF A FEATHER?

I ate Korean chicken for lunch yestereday and ate Japanese chicken for lunch today. That made me wonder about the relationship, if any, between Middle Korean tʌrk 'chicken' and Old Japanese təri 'bird'. I was surprised that I could not find this pair in Vovin's 2011 Koreo-Japonica. It is in the database of the late Sergei Starostin:

Starostin's "Korean" (sans prefix) reconstruction is based solely on Korean and not other early peninsular languages, so strictly speaking it is Proto-Korean rather than Proto-Koreanic.

I have made some minor orthographic changes: e.g, *th instead of *t` for the PA aspirate.

Although one might initially want to reconstruct initial *t-, Starostin reconstructed aspirated PA *th- for the correspondence set

PTurk *t : PM *t : PTun *t : PK *t : PJ *t

whereas he reconstructed unaspirated PA *t- for the correspondence set

PTurk *d : PM *d : PTun *d : PK *t : PJ *t

The problems with the etymology start with the first vowel. PA initial *thŏ́- in 'net' corresponds to

PM *to (not *tu as in 'kind of bird'!) and

PJ *tú (possibly from *tó, not *tə́ as in 'kind of bird'!)

Which set of correspondences is regular?

If one looks at correspondences of PA *o in general, there is even more apparent chaos: e.g.,

PA *tho- corresponds to PTun *to-, not *tu- in 'fire'

PA *thŏ̀- corresponds to Starostin's PK *thó-, not *tʌ̀- in 'hare' (which I think is a loanword from Chinese 兔 and has nothing to do with Altaic)

PA *thò- corresponds to PJ *tà-, not *tə̀- or ́*tù- in 'to see, beware'.

Such a mess points to loanwords (e.g., 'hare') and/or lookalikes, not true cognates, unless one reconstructs more proto-vowels to account for the variety of reflexes.

The second vowel is puzzling. How did Starostin reconstruct *o as the source of *a-vowels in Mongolic and Tungusic and *i in Japonic? Old Turkic has yet another vowel ɨ in torɨɣa. What if there were an area word (originally from Xiongnu?) with an original cluster *rk that was broken up with different vowels in the various Altaic languages?

The voicing of the consonant after the second vowel is not consistent. I would expect *g instead of *k in Proto-Tungusic. 'Wrinkle' has yet another pattern:

The vowels after that consonant do not match at all within each correspondence set.

My guess is that

- The Mongolic and Tungusic words for 'crow' are related, and the PM *ɣ : PTun *k mismatch points to borrowing (i.e., inaccurate replication). The Turkic word may also be related. Perhaps the original meaning was 'dark bird' (crows are black and larks are brown).

- The Japonic word for 'bird' is a borrowing from Koreanic.

Japonic had no *ʌ, so Koreanic *ʌ was borrowed as Japonic *ə.

Japonic *i might reflect a lost *i in Koreanic between *r and *k.

That *-i might also be the original final vowel if the Koreanic word were *tʌri at the time of borrowing; a suffix *-k(V) could have been added later (but what would it have meant?).

The Koreanic word may ultimately be from the same source as the western Altaic words.

Martin (1992: 98) suggested Old Chinese 鳥 *tyeg 'bird' as a possible source of the Koreanic word for 'chicken'. However, there is little resemblance between my Old Chinese reconstruction *Cʌ-tiwʔ and Middle Korean tʌ̀rk. The OC presyllable could have been lost at the time of borrowing into Koreanic, but the vowels do not match.

It is not clear how OC *Cʌ-tiwʔ is related to non-Chinese Sino-Tibetan words for 'bird'. Maybe the presyllable began with an *n- (still in Mandarin niao 'bird') that fused with *t-, resulting in the d- of Bodo-Garo and Liangshan Yi. (But if *nt- became d- in Liangshan Yi, where did Liangshan Yi nd- come from? *nd-? A later layer of fusion?) The vowel variation in Sino-Tibetan may reflect ablaut. The Proto-Sino-Tibetan root could be *t-wʔ.

12.12.16.23:59: THE ROOT OF THE SAOSHYANT

Shortly after I first started learning Sanskrit, I found Jackson's An Avesta Grammar in Comparison with Sanskrit which made me wonder what the Sanskrit cognate of Avestan saošyaṇt- 'Saoshyant' was or would be. Theoretically it should correspond to a Sanskrit *śoṣyant-, a future participle of a root √śū.

Tonight I discovered that Skjærvø (2006: 29) defined the root of saošyaṇt- as 'revitalize' and saw it in Avestan asūra- 'not having vitalizing power', which he compared to Sanskrit śū́ra- 'strong'. (a- is a prefix 'not' and -ra is a suffix in both languages.)

So I guess the cognate Sanskrit root is indeed √śū which also has variants śvā and śvi. Could these all be from a Proto-Indo-European *√k-w-H?

zero grade: *k-w-H > *k-u-H > Skt śū, Av sū

alternate zero grade: *k-w-H > Skt śvi (cf. PIE *pʕter- > Skt pitar- 'father' - but how did *ʕ become i instead of a?; Avestan has ptar- [later patar-] without any vowel between p and t)

e-grade: *k-e-w-H > Skt śo, Av sao

alternate e-grade: *k-w-e-H > Skt śvā

In theory Proto-Indo-European *√k-w-H could have contrasted with a root *√kʷ-H, though I don't know of any such root.

It occurred to me that if PIE *k were as common as other voiceless stops, its Sanskrit reflex ś should be as common as Skt t and p. Yet according to Whitney's (1896: 26) statistics, Skt ś from PIE *k is much less common than Skt t and is even less common than the reflexes of PIE *kʷ (Skt k and c) combined. Was PIE *k less common than PIE *kʷ even though the latter is more marked? (In Cantonese, /kʷ/ is less common than /k/.)

12.12.17.00:17: The resemblance of Greek σωτήρ sōtḗr 'savior' to saošyaṇt-, the final savior in Zoroastrianism, is coincidental. True Greek cognates have k-: e.g., ἄκυρος ákyros 'without authority' (directly corresponding to Av asūra- 'not having vitalizing power'?) and κύριος kýrios 'having authority' (nearly corresponding to Skt śū́ra- 'strong').

12.12.17.00:34: Hypothetical lookalikes of σωτήρ with reflexes of the same PIE suffix are Avestan *saotar- and Skt *śotar-.