There's no final -t in the title unlike last night's post.

a is probably the most common vowel of all (though Proto-Indo-European has been reconstructed without it!). Short a alone constitutes one out of every five sounds in a Sanskrit text (Whitney 1889: 26). So I was initially surprised to read in Short (1993: 456) that in Czech,

Initial a-, e-, and i- only occur in loan-words, the conjunctions a and i 'and', ale 'but', and some interjections.

Then I remembered that Proto-Slavic 'grew' an initial *j- where there was none, resulting in Russian яблоко jabloko corresponding to English apple (without initial y-).

According to Schenker (1993: 68), *j- developed before front vowels (*i, *ii, *e, *ee) and long *aa- in Proto-Slavic. *j is a consonantal counterpart of the front vowel i, so I'm not surprised that it appeared in front of front vowels. However, Schenker (1993: 66) reconstructed *aa as a back vowel ([ɑɑ]?). Why would j- be added to a low back vowel? The other long back vowel in Schenker's reconstruction (*uu) shifted to *wuu, not *juu. WHAT'S UP WITH YAT?

Serbo-Croatian dialects can be classified according to the vowels they have in certain words. Ekavian dialects have e, ikavian dialects have i, and ijekavian dialects have (i)je:

'river' 'faith'
Ekavian reka vera
Ikavian rika vira
Ijekavian rijeka vjera (not vijera)

Some dialects have an upper mid vowel between mid e and high i. I will write this vowel as é and call these dialects ékavian. One might guess that Proto-Serbo-Croatian, the common ancestor of all these dialects, had an é which remained in some dialects but developed into e, i, or (i)je elsewhere.

However, these vowels descend from the Proto-Slavic (PS) vowel yat (ѣ = [æ]) which was low, not upper mid.

Vowel height Vowel
High i
Upper mid é
Mid e
Low æ

PS in turn has two types of sources:

earlier long *ee

earlier *aj and *aaj

The merger of with the slightly higher vowels e in ekavian is understandable. But how did end up becoming much higher vowels or diphthongs (é, (i)je, i) in other dialects? Simple raising cannot be the answer. If raised to e on the way to é or i, then speakers would have no idea

which e were from

i.e., which e were eligible for further raising

which e were original

i.e., which e were not eligible for further raising

I am not a Slavicist, so I don't know what happened. I can only guess that in non-ekavian dialects, broke to something like *jæ. Cf. the development of *-j- before in Middle Chinese and the substitution of ya [ja] for [æ] before k and g in English loanwords in Japanese.

In non-ekavian dialects, *jæ shifted to (i)je.

In ijekavian dialects, (i)je remains intact.

In ékavian dialects, (i)je fused into é.

In ikavian dialects, (i)je simplified to i (or fused into é which then raised to i)

If ékavian and/or ikavian dialects have words with (i)je

- this scenario is wrong (as it predicts that all (i)je would shift to é or i)

- or those words may have environments in which (i)je is preserved

- or those words are loans postdating the changes involving (i)je TANGUT '(K)NEW'

Last week, I mentioned two cognates without noting that they were related:

Mawo Qiang khsə 'new'

Tangut siw 'new' < pre-Tangut *sik


Taoping Qiang tshi 'new' < ?*ksi

Japhug rGyalrong kɯ ɕɤɣ 'new' (a loan from an eastern rGyalrong form similar to Somang kə ɕə́k? < Proto-rGyalrong *-ək?)

Written Burmese sac 'new' < *sik

Old Chinese 新 *sin 'new' < ?*siŋ < *??sik-N (with a final nasal instead of a stop)

Proto-Tibeto-Burman *g-sik (Matisoff 2003: 344)

Earlier, we saw that both MQ and Tangut shifted the *i of 'two' to a central vowel unlike Taoping Qiang. But this time only.the MQ word has undergone that shift, whereas Tangut siw retains the earlier *i like Taoping Qiang. Like MQ, Proto-rGyalrong has a instead of an *i.

Strict regularity would require two different vowels for 'two' and 'new' in the common ancestor of the rGyalrong, Qiang, and Tangut languages, but I suspect both words shared a single vowel. What accounts for their different developments?

1.6.1:57: Japhug rGyalrong kɯ ɕɤɣ seems to preserve an old *kV- presyllable that has developed differently elsewhere:

*kV- > kh- before -s- in MQ khsə

*kVs- > khs- > tsh- in Taoping Qiang tshi

*kV- was lost entirely in Tangut siw, if it ever existed in the Tangut line; the high vowel of siw implies that any presyllable also had a high vowel (like the kɯ- of Japhug)

This *kV-presyllable (corresponding to Matisoff's PTB *g-) may not have been in all Sino-Tibetan branches. If it ever existed in the Burmese or Chinese lines, it disappeared without a trace.

The root initial ɕ- in Somang kə ɕə́k (and Japhug kɯ ɕɤɣ, a loan from eastern rGyalrong) is odd because Proto-rGyalrong *s- normally palatalizes to ɕ- before front vowels in Somang. Perhaps

PG *sik > *ɕik > Somang ɕək

But what would cause *i in 'new' to shift to ə in Somang (or in MQ)? NO, DEAR, IT'S A TIGER

On Saturday, I speculated that

the left and right component of the Tangut character for


might be based on 乕, a variant of Chinese 虎 'tiger'. But what is the function of the middle element

dzie 'one of a pair'

which has no phonetic or semantic similarity to Tangut ləi 'tiger'?

While looking at other variants of Chinese 虎 'tiger', I discovered some that have an X-like element (cf. 'one of a pair' above) flanked by E-shaped elements that are mirror images of each other. This EXヨ pattern occurs beneath a top element: e.g., 鹿 'deer' in this variant. If

- the top element is removed

- the Es are slightly altered and face the same direction

- the X becomes more elaborate

the result is the tangraph for 'tiger'. Is the similarity between these EXヨ variants and Tangut 'tiger' coincidental? Were any of the EXヨ variants used in northwestern China prior to the invention of the Tangut script?

Was the inventor of the Tangut script familiar with Shuowen? Shuowen lists two variants for 虎 'tiger'. One is of the EXヨ type and the other looks like 鹿 'deer' + 勿 'do not' which do not have any obvious semantic or phonetic similarity to 'tiger'. Could these variants be distortions of a drawing of a tiger?

The standard graph for 'tiger', originally a drawing of a tiger, now appears to contain 七 'seven' and a pair of 儿 legs (which can also be a simplification of 兒 'son'). The 'seven' seems to be a distortion of an earlier pair of legs. Karlgren (1957: 45) regarded the 'legs' as a distortion of a tiger's tail. (But why would a tail fork? Could one half of 儿 have been a hind leg?) THIS YEAR'S TANGUT NUMERALS: NIƏƏ

I am low on energy but still want to post something each day, even if it's as short as this entry.

The Tangut numeral for 'two'

was transcribed into Tibetan as gni(H) and nyi and into Chinese as (o)能 (pronounced something like nə̃ in Tangut period northwestern Chinese?). Gong reconstructs it as njɨɨ and I reconstruct it as niəə.

The word belongs to rhyme 33 which has various Tibetan transcriptions: usually -i(H), but also -a and there is one instance each of -ang with an unexpected nasal coda and -e. Tibetan has no letters for ɨ or ə, so i ~ e ~ a could be attempts to write such vowels..

Tangut period northwestern Chinese probably had no syllables like niəə or even nəə or nə, so 能 nə̃ was the closest available approximation of Tangut niəə.

My niəə is not far from Mawo Qiang ɣnə. But those two forms have a schwa unlike

Taoping Qiang ɲi

Written Tibetan gnyis (the g- corresponds to Mawo Qiang ɣ-)

Written Burmese hnac < *hnit

Old Chinese *nis (there is no Chinese-internal evidence for *g-)

Matisoff (2003: 135) reconstructed Proto-Tibeto-Burman *g-nis 'two'. I am not yet convinced that PTB even existed (i.e., that non-Chinese Sino-Tibetan (NCST) languages form a single branch), but I cite his reconstruction to indicate that NCST evidence presumably points to *i as the vowel for 'two'. (I say presumably because Matisoff's knowledge of NCST dwarfs mine by orders of magnitude.)

One could reconstruct a Proto-Sino-Tibetan diphthong like *iə that

- was more or less preserved in Tangut

- simplified to ə in Mawo Qiang

- simplified to i elsewhere

but I do not like reconstructing proto-phonemes on the basis of a single word and I seriously doubt Tangut preserved a diphthong lost in all other ST languages. I suspect that Tangut and Mawo Qiang 'neutralized' *i to (but why? vowel harmony with a presyllabic vowel?) and that the Tangut -i- reflects an earlier presyllabic high vowel, not PST *i:

PST *kɯ-nis > Pre-Tangut *kɯ-nəə > *kɯ-nɨəə > Tangut niəə

I don't understand why the Tangut forms have long vowels and have a 'level' tone instead of the 'rising' tone via the pre-Tangut *-H that I would expect from PST *-s.

Tangut fonts by Mojikyo.org
Tangut radical font by Andrew West
All other content copyright © 2002-2010 Amritavision