There are only three tangraphs with the left-hand element vem (in David Boxenhorn's code):

0200 2lõ 'relative'

0642 2lõ 'origin'

1018 1lwo 'moist'

I could call this trio of lo-characters a vem-śa (vaṃśa- being Sanskrit for 'family').

Only the third has a known Tangraphic Sea analysis:


1018 1lwo 'moist' =

left of 0642 2lõ 'origin'(phonetic)

+ center of 3052 1niooʳ 'water (trigram ☵)' (semantic)

It is clearly a phonosemantic compound. I should look into how many tangraphs with oral vowel readings have nasal vowel phonetics and vice versa.* I am also interested in how many tangraphs with -w-readings have -w-less phonetics and vice versa. The looseness of the 'fit' of Tangut phonetics has yet to be measured.

I presume 'relative' is phonetic in 'origin' or the other way around. It is also possible that 'origin' is semantic in 'relatives' (i.e., people sharing a common origin).

The right side of 0200 (Boxenhorn code: vemqii)

is semantic; it is shared with ten other tangraphs:

Tangraph LFW2008 Boxenhorn code Reading Gloss
0199 fulqii 2siə first half of 2siə 2sa 'to connect' (i.e., to make near)
0213 fioqii 1nie relative (people who are near to one)
0915 qiitos 1thaa to haunt, make mischief (to be haunted is to have ghosts [symbolized by the right element tos] nearby, and ghosts make mischief)
1639 qiibaehae 1khwạ far (opposite of near)
1951 ciadexqii 2sa Second half of 2siə 2sa 'to connect'
1957 gaaqii variant of 0213
2217 dexbelqii 1ɣʌ near; not attested outside dictionaries? 'ritual' word? resemblance to Middle Chinese 近 *gɨnˀ 'near' coincidental?
2223 bilhasqii 2rieʳ to mend, sew (to seal holes in clothing is to connect threads - to make them near each other)
4506 banqiiqex 2ləu to burn, ignite, light (the right element is qex 'fire')
5228 haehasqii ?lẽ husband of sister (relative and hence near)

Nearly all have glosses involving nearness (0199, 0213 = 1957, 0915, 1639, 1951, 2217, 2223, 5228). The sole exception is 4506 in which qii might be phonetic: 2ləu is vaguely like 2lõ, the reading of 0200.

Unfortunately the right side of 0642 (Boxenhorn code: vemjolcon)

is unique. It may be a compound of parts (jol and con) extracted from two tangraphs: one of 15 others with jol and one of 12 others with con.

*Gong reconstructed rhyme 2.47 as -ow which corresponds to my -õ. If Gong was correct, vem represented lo-like syllables with or without medial or final glides.

In any case, I don't know why only three lo-like syllables were written with vem while 41 others were not. "V NANGOMU"

I am puzzled by Nangomu in the title of this article in a Slovene newspaper: "Trimesečna pomoč v bolnišnici v Nangomu" 'Three months' aid in a hospital in Nangoma'. -a nouns are typically feminine in Slavic. The preposition v 'in' takes the locative case*, and Nangomu is the locative singular of Nangoma as if it were ... masculine? However, the body of the article contains what I'd expect: "v Nangomi" with the locative singular of a feminine Nangoma.

Can foreign place names ending in -a be declined in Slovene as if they belonged to either gender? On .si sites**, Google has 523 results for "v Nangomi" (f.) but only 5 results for "v Nangomu" (m.). Similarly, Google has 357,000 results for v Ameriki" (f.) but only 375 results for "v Ameriku" (m.) including Russian transliterations (so the real figure is smaller). It's not surprising that Wiktionary lists Slovene Amerika as a feminine noun. Should Nangoma also be solely feminine? Most Slovene -a nouns are feminine, not masculine. Why decline Nangoma according to a minority pattern? Is there a Nangoma-like masculine -a noun motivating an analogical declension? Compare how English dive has an irregular conjugation by analogy with the rhyming verb drive.

Returning to the Slavic world, I have never understood how even recent loanwords in Russian came to have the less common -a ending in the plural: e.g., svitera as well as svitery 'sweaters' with the usual -y ending. Isn't this -a an old dual ending: e.g., in glaza 'eyes' (formerly 'two eyes')? I can understand why -a became the plural ending for paired items but not how it spread to nouns not strongly associated with 'two': e.g., professora 'professors'.

*V can also take the accusative case when it means 'into', but 'into' makes no sense in the title.

**The article I quoted is an .eu site. If I search through all Slovene sites, Google will not display the number of results. I can go to the end of the results and see a number, but that figure excludes similar entries and can't be compared to the results for .si sites which do include similar entries. Including similar entries in a Slovene search results in no numbers at all - not even if I go to the last page. <X.Ü.N M.IN TZ.ING I.M>

On Sunday I discovered a virtual 訓民正音 Hunmin chŏngŭm online and wished that equivalent documents could be found for the Tangut, Jurchen, and Khitan scripts. Although I agree with Juha Janhunen' s (1994, 1996) hypothesis of the Khitan and Jurchen large scripts being the results of evolution rather than invention, the Khitan and Jurchen small scripts and the Tangut script were all conscious inventions, and in theory their creators could have written manifestos. But in reality ... who knows?

Although the Khitan small script is believed to have been invented around 925, the earliest known example of the script (耶律宗教 Yelü Zongjiao's epitaph) is from 1053. Compare that 128-year gap with the three-year gap between the invention of hangul in late 1443 or early 1444 and the publication of Hunmin chŏngŭm in October 1446. For fun I've written the Sino-Khitan pronunciation of 訓民正音 Hunmin chŏngŭm below in the Khitan small script:

<x.ü.n m.in tz.ing i.m>

The Jurchen small script is believed to have been invented in 1138. Only a handful of undeciphered potential examples (e.g., the block on the right) survive. Aisin Gioro Ulhicun theorized that the script died out following the assassination of its inventor Emperor Xizong.

The Tangut script is believed to have been invented around 1036. Modern scholars have tried to reverse-engineer the script with the aid of the Tangraphic Sea's analyses, but the latter are circular and unreliable. Hence the structure of tangraphy remains only partly understood. Although all agree that there is a large number of recurring components ('radicals'), no native source explictly names these radicals, much less states what their functions are*. I would not go as far as Janhunen (1994: 10), who wrote,

The attempts made so far to analyze the Tangut 'characters' in terms of a 'radical' structure of the Chinese type are so unconvincing that they only corroborate the impression that the Tangut script cannot have been of the 'ideographic' type.

However, I admit that Tangut radicals are often seemingly arbitrary without any obvious semantic or phonetic function: e.g., why do nearly one out of five tangraphs** have Nishida's radical 204

which is often called 'person' though it often appears in tangraphs that have nothing to do with 2dzwio 'person'?

1nɨaa 'black'***, 1na 'dog', 1giəə 'nine'**** - three of the 1,187+ tangraphs with 'person'

A Tangut Hunmin chŏngŭm might clear this up - though experience has taught me that more data means more mysteries in Tangutology.

*The Tangraphic Sea describes tangraphs in terms of the left, right, etc. of other tangraphs; it only implies the existence of radicals without actually including those radicals in its text. Amazingly no dictionary in the vast native lexicographic tradition is organized by radical. Is that absence telling?

**This figure excludes tangraphs with radicals incorporating the shape of 'person': e.g.,

Nishida's radical 302 / 2to 'end'

This tangraph is <demavixe> in Downes' (2008) transliteration system; Nishida's radical 204 / 2dzwio 'person' is <xe>. Downes (2008: 20) transliteration

is designed to act as a mnemonic to aid in remembering the structure of individual characters - it has no relation to how the language behind the Xixia script may once have sounded.

The graphic relationship between Nishida's radicals 302 and 204 is also visible in their stroke order codes in this Unicode Tangut proposal: DCGQCCCQ and CCCQ.

***The Tangraphic Sea analysis for 'black' is incomplete and circular:


1nɨaa 'black' = ? + right of 1kõ 'night' (which is obviously 'black' plus the mystery component 'person')

The missing first tangraph might be 1na 'dog' which would be phonetic.

1nɨaa 'black' is phonetic in 1na 'dog':


1na 'dog' = left of 1khwiə 'dog' + left and center of 1nɨaa 'black'

1khwiə has an unexpected central vowel; its cognates have front vowels: e.g., Written Tibetan khyi, Written Burmese khveḥ, and Middle Chinese 犬 *kʰwenˀ (whose Old Chinese source is unclear). Perhaps the original vowel was *ɨ.

One might think that the shared components of 1nɨaa 'black' and 1na 'dog' constitute a tangraph with a reading like na, but in fact

'how, where'

was pronounced 2lɨọ! (Is 'person' on its right due to the influence of the sinograph 何 'what' which has 亻 'person' on its left?)

****Is 'person' in the tangraph for 'nine' due to the influence of the sinograph 仇 which has 亻 'person' on its left and was almost homophonous with 九 'nine' in Tangut period northwestern Chinese?

Tangut fonts by Mojikyo.org
Tangut radical and Khitan fonts by Andrew West
Jurchen font by Jason Glavy
All other content copyright © 2002-2013 Amritavision