When I discovered Ostapirat's (2000) Proto-Kra back in April, I was puzzled by his reconstruction of Proto-Kra retroflex consonants. Initially I wanted to reconstruct *Cr-clusters, but Ostapirat already reconstructed such clusters and his retroflexes lacked retroflex or even rhotic reflexes in the modern languages. Later, I found Guillaume Jacques' (2001) review of Ostapirat which also expressed concern over retroflex stops and laterals* whose reflects are "identical to the dental series in all branches but Paha" (p. 103). I think Guillaume's solution is correct: Paha ð(h)- is the result of presyllable and cluster simplification. I suspect that Paha ð(h)- is parallel to Paha v- and ɣ-:

Paha v- < Proto-Kra *C-p-

Paha ɣ- < Proto-Kra *C-k-

hence Paha ð(h)- < Proto-Kra *C- + dental?

This lenition is in turn parallel to lenition in Vietnamese:

v- < [β] < *C-p-/*C-b-

d- [z] < [dʲ] (could Middle Vietnamese d- have been [ð]?) < *C-t-/*C-d-

gi- [z] < [dʑ] (could Middle Vietnamese gi- have been [ʑ]?) < *C-c-/*C-ɟ-

g- [ɣ] < *C-k-/*C-g-

In Hanoi, lenited dentals, lenited palatals, and *r- have merged as [z]. Similarly, in Paha, lenited dentals and *(h)r- have merged as [ð].

Paha lenition is also reminiscent of my proposal for lenition in Tangut:

w- < *C-p(h)-/*C-b-

l- < *C-t(h)-/*C-d-

z- < *C-(t)s(h)-/*C-(d)z-

ʑ- < *C-(t)ɕ(h)-/*C-(d)ʑ-

ɣ- < *C-k(h)-/*C-g-

*Guiillaume did not discuss Ostapirat's Proto-Kra retroflex affricates *tʂ and *dʐ. Unlike the other retroflexes, these two consonants did not become ð in Paha. I will examine them in my next post. CORRESPONDENCES BETWEEN PROTO-KRA AND PROTO-TAI IMPLOSIVES

Now I'll move on to another branch of Kra-Dai. What do Proto-Tai implosives (or preglottalized stops) correspond to in Ostapirat's (2000) Proto-Kra reconstruction? I compiled the following table using Ostapirat's list of Kra-Siamese cognates (pp. 3-8), his Proto-Kra word list (pp. 219-245), and Doug Cooper's Proto-Tai'o'Matic:


Glosses of examples
























Proto-Kra b, d, ɖ are "better recognized as implosives" despite their notation: [ɓ ɗ ᶑ] (Ostapirat 2000: 209). According to Wikipedia,

Retroflex implosives (e.g., [ᶑ]) have not been confirmed to exist in any language, though one has been claimed for Ngad'a, an Austronesian language spoken in Flores.

I would like to derive the exotic Proto-Kra implosive [ᶑ] from a Proto-Kra-Dai cluster:






*C-b- or *C-p-



b2, b3

*C-br(j)- or *C-pr(j)-




*C-dr- or *C-tr-




*C-dl- or *C-tl-


A PKD *C- eroded to *ʔ- and fused with following stops, resulting in PK and PT implosives: e.g.,

PKD *C-b- > *ʔb- > PK *ɓ-

PK *m- and *k- could have been added later to roots which had developed implosive initials.

PK retroflex implosives are from PKD clusters with *-r- (b2, d1). In b2, the labial stop assimilated to the following liquid in pre-PK:

PKD *C-br- > *ʔdr- > PK *ɖ- [ᶑ]

I reconstruct PKD *-l- in d2 because PK has nonretroflex *d-. HEITU AND TONGSHI FRICATIVES CORRESPONDING TO PROTO-TAI *X AND

If Heitu and Tongshi g is not from *g, what did *g become? I assume that Proto-Hlai *g corresponds to Proto-Tai *g, but I can't find any PT *g-words in the comparisons in Ostapirat (2000: 63-65). However, I did find Heitu and Tongshi cognates for at least two PT -words:

Correspondence Glosses of examples Heitu Hlai Tongshi Hlai Siamese Proto-Tai
x1 bitter h h kh *x/ɣ
x2 excrement z
x3 crow
x4 cogon, gold

Ostapirat's Siamese tones imply that 'bitter' and excrement' could have had PT *ɣ- as well as *x-. (Only a voiceless initial is reconstructible for those words in Li Fang-kuei's system, and the modern spellings for the Siamese forms imply voiceless proto-initials: ขม and ขี้. The rising tone of 'bitter' in Siamese unambiguously points toward a voiceless proto-initial, though the falling tone of 'excrement' is ambiguous.)

The tones of all of the above Heitu and Tongshi forms point to proto-voiceless initials, even though Tongshi z- is voiced.

The correspondence of Heitu h- to Tongshi z- is reminiscent of the correspondence of Sanskrit h- [ɦ] to Avestan z-. Just as Avestan z- came from Proto-Indo-European *ghʲ-, Tongshi z- could have been from something like *ʑ- < *ɣʲ- < *ɣ-. Some instances* of Spanish j [x] may be the result of a similar change in reverse:

[x] < *xʲ < *ɣʲ- < *ʑ- < *j

However, Tongshi z- also corresponds to PT *x- in 'crow'. I would have expected 'crow' to have h- like the Tongshi word for 'bitter'. Is 'crow' exceptional because it is onomatopoetic?

I have long been puzzled by PT *ɣam 'gold'. It looks like Old Chinese *kəm 'gold' but the initials don't match. Could the PT initial reflect a southern OC dialect form with *ɣ- from root-initial *k- plus some prefix(es)? (Prefix-conditioned lenition occurred in Vietnamese centuries later: Viet g [ɣ] is from *C-k- and *C-g-.)

*Another source of Spanish [x] is *s:

[x] < *xʲ < < < *s

e.g., in jugo 'juice' < *suuku "via a medieval pronunciation of /s/ as [ɕ] in the southern dialect of Mozarabic" (Green in Comrie 1987: 239). HEITU AND TONGSHI G

Some HIai languages (Baoding, Heitu, Tongshi) have a g which does not correspond to Proto-Tai *g. Ostapirat (2000: 63-65) provided the following comparisons with Siamese:

Correspondence Glosses of examples Heitu Hlai Tongshi Hlai Siamese Proto-Tai
g1 bran, we r g r *r
g2 head *kl
g3 ask g th *thl
g4 taro (Baoding g) ? ph *phl/r

Heitu and Baoding tones imply that the initial of g1-g4 was once voiceless *hr-.

A chain shift could have occurred in the common ancestor of Heitu and Baoding:

Step 1: *k-r- (did 'bran' and 'we' have a Proto-Hlai velar not in Proto-Tai?), *kl- > *hr- > r-

Step 2: *thl-, *phl/r- > *hr- (after earlier *hr- was lost in step 1) > *ʀ- > *ʁ- > *ɣ- > g-

In Tongshi, perhaps
*voiceless stop + l/r- > *hr- > *ʀ- > *ʁ- > *ɣ- > g-

with one reflex for both *kl/r- and *thl-. (7.24.0:02: I would expect *pl/r- to also become Tongshi g-.) IMPLOSIVES ON HAINAN: HEITU AND TONGSHI NONIMPLOSIVES CORRESPONDING TO TAI IMPLOSIVES

I compiled the table below using data from Ostapirat (2000: 63-65). Siamese d is from an earlier *ɗ.

Correspondence Glosses of examples Heitu Hlai Tongshi Hlai Siamese Proto-Tai
ɓ1 shoulder v f b
ɗ5 thunder/crow (v.) r r d ?
ɗ6 bone, raw f *ɗl/r
ɗ7 navel *ɓl/r
ɗ8 moon ɲ ɲ

ɓ1: Could Heitu and Tongshi partly preserve an earlier *ʔv- that merged with *ʔp- and became in Proto-Tai? The Heitu and Tongshi words for 'shoulder' have tone 2 implying an earlier (Proto-Hlai?) voiced initial *v- (simplified from Proto-Kra-Dai *ʔv-?).

ɗ5: The Proto-Kra-Dai initial could have been *ʔr-.

ɗ6 and ɗ7: Heitu and Tongshi have lost a distinction maintained in Proto-Tai:

In Heitu, ?*ɓl/r- and *ɗl/r- lost their first element and became r-.

In Tongshi, ?*ɗl/r- dissimilated to ?*ɓl/r- to avoid a double alveolar cluster and then became f-.

ɗ8: The Proto-Kra-Dai initial could have been *ʔml-. Cf. Vietnamese nhẽ ~ lẽ 'reason' from Middle Vietnamese mnhẽ ~ mlẽ (in turn from a Late Old Chinese ?*mlɛh < Old Chinese *mʌ-rəʔ-s; note that 理 'reason' has the same phonetic as the m-graph 埋 'bury').

A nasal element in the initial is difficult to reconcile with Proto-Austronesian *bulaN 'moon' which is presumably somehow connected to the Kra-Dai word.

Perhaps Proto-Kra-Dai *ʔbl- became Proto-Hlai *ʔml- but merged with *ʔpl- to become Proto-Tai *ɓl/r-:

Proto-Kra-Dai Proto-Hlai Proto-Tai
*ʔbl- *ʔml- *ɓl/r-

I compiled the table below using data from Ostapirat (2000: 63-65). I assume Ostapirat's Heitu d is [ɗ] or at least from an earlier (like Siamese d).

Correspondence Glosses of examples Heitu Hlai Tongshi Hlai Siamese Proto-Tai
ɗ1 nose/face d d d
ɗ2 gall bladder *ɓl/r
ɗ3 fish ɬ pl *pl
ɗ4 tongue, blood, deep l *l

ɗ1 is straightforward and points toward Proto-Kra-Dai *ɗ- < *ʔt-.

d in ɗ2 may result from parallel, independent simplifications of a PKD cluster:

*ʔpl/r- > *ɓl/r- > *ɗl/r- > d-

Maybe the simplification had already taken place in Proto-Hlai, whereas its sister Proto-Kam-Tai retained a cluster that was passed down to Proto-Tai.

The PKD initial of ɗ3 may have been *pl-. This could have been reduced to *ʔl- in Proto-Hlai which became Heitu d and Tongshi ɬ (via *hl?).

The PKD initial of ɗ4 may have been *ʔl- which merged with secondary *ʔl- < *pl- in Proto-Hlai and with *l- in Proto-Tai. IMPLOSIVES ON HAINAN: BAODING IMPLOSIVES CORRESPONDING TO TAI NONIMPLOSIVES

I compiled the table below using data from Shintani (1991: 2). The numbering of correspondences continues from the table in the previous post.

Correspondence Common Tai Zhuang Be Baoding Hlai Jiamao Hlai
ɓ8 筆 *p- > ɓ- (none) p ɓ ɓ ɓ
ɓ9 變 *pr- > *p- > ɓ- pl
ɓ10 f f f, ɓ
ɓ11 浮 *b- > *v- > *f- v f, p v ?
ɗ7 t t ɗ ɗ ɬ
ɗ8 凳 *t- > ɗ- ?
ɗ9 th
ɗ10 銅 *d- > *t- > ɗ- d h t, ɗ
ɗ11 h ɣ ?
ɗ12 ? ? t
ɗ13 茶 > *t- > ɗ- ʃ s
ɗ14 l ɗ l
ɗ15 ʔj ʔj j tsh

In six cases (ɓ8-9, ɓ11, ɗ8, ɗ10, ɗ13), the unusual correspondences may be due to borrowings from different dialects of Chinese at different periods.

The remaining cases presumably involve native words and therefore require internal explanations:

ɓ10: Could the Proto-Kra-Dai initial have been *ʔf- (from an earlier *ʔVf-; I assume that all glottal stop clusters are reductions of earlier *ʔV-).

ɗ7: 'castrate' has Proto-Tai *t- but the Proto-Kra-Dai initial could have been something more complex like *ʔtɬ- (if the Jiamao lateral is conservative).

ɗ9: 'arrive' has Proto-Tai *t- but the Proto-Kra-Dai initial could have been something more complex like *ʔtɬh-, the aspirated counterpart of the ɗ7 initial. (But why isn't the Zhuang initial aspirated?)

ɗ11: 'mushroom' has Proto-Tai *hr-. Could the Proto-Kra-Dai initial have been *ʔr-?

ɗ12 involves a word absent from the non-Hainan languages. The Be form is presumably a loan from a Hlai language: *ʔr- borrowed as *hr- > h-?

ɗ14: I have no idea why Zhuang has an implosive instead of a liquid. The Proto-Kra-Dai initial might have been something like *ʔ(t)l-.

ɗ15: Did *ʔj- harden to Baoding ɗ- and Jiamao tsh-? Are there any examples of Tai ʔj- corresponding to Hlai j-?

The Tai (ʔ)j- : Jiamao tsh- correspondence is reminiscent of a few Chinese phonetic series listed in Schuessler (2007: 97):

酉 Md you : 酒 Md jiu < *ts-

亦 Md yi : 跡 Md ji < *ts-

允 Md yun : 㕙 Md qun < *tsh-, jun < *ts-

The Chinese affricates may reflect lost presyllables: e.g., *sɯ-tɯ-j- > *st-j- > *tsj-. Could Jiamao tsh- also come from presyllables or prefixes plus *(ʔ/h)j-?

Tangut fonts by Mojikyo.org
Tangut radical font by Andrew West
All other content copyright © 2002-2008 Amritavision