Home

15.11.30.23:11: HOW NOT TO RECONSTRUCT MON-IC RHYMES

(Posted after expansion on 15.12.18.)

I thought it might be fun to try to work out my own reconstruction of Monic rhymes using the data in Diffloth (1984: 286) before seeing his solution. Real forms are in bold.

Correspondence 1 2 3 4 5 6
Proto-Monic *-eK *-eC *-iC *-iK *-ɛC *-ɛK
Pre-Nyah Kur *-iK *-iC *-iiC *-iiK *-ɛeC *-ɛeK
Nyah Kur -iC -iiC -iiK -iiC -eeC
Pre-Literary Mon *-eK *-iC *-iK *-eC *-eK
Literary Mon <-eK> <-iK> <-eK>

I assumed that Mon generally preserved vowel heights: <e> rhymes came from Proto-Monic nonhigh vowels, and the vowel of *-eC raised to assimilate to the following palatal.

My solution has a short absent from the reconstruction in my last post. (But such a vowel was in an earlier draft of that post. Both versions of my reconstruction lacked length.)

Capital letters stand for palatals (-c, -ɲ) and velars (-k, -ŋ).

I posit a chain shift in Nyah Kur:

1. *iii (contra last night's post in which *e > *ei > Nyah Kur ii)

2. *e > i (and *-iK > *-iC - but why didn't -iiK become *-iiC?)

3. > *ɛe

3a. *-ɛeC > *-eiC > -iiC

3b. *-ɛeK > *-eeK > *-eeC

In Written Mon, lower mid front vowels raised, and palatals backed:

1, *-eC > *-iC

2. > <e>

3. *C > <K>

15.12.18.11:22: I finally got around to looking at Diffloth's solution two and a half weeks later. He used Old Mon and non-Monic languages to help him. I wonder what my solution would have looked like if I had checked that data..

Correspondence 1 2 3 4 5 6
Proto-Monic (this site) *-eK *-eC *-iC *-iK *-ɛC *-ɛK
Proto-Monic (Diffloth 1984: 288) *-iK *-iC *-iiK *-eeK *-iiC *-eeC
Pre-Nyah Kur (this site) *-iC *-iiC *-eeK *-iiC *-eeC
Nyah Kur -iC -iiC -iiK -iiC -eeC
Pre-Old Mon (this site) *-eC *-iC? *-iiK *-eeC
Old Mon (phonetic; this site) *-eiC *-iC? *-i(i)K *-eiC
Old Mon <iC> (once) ? <-iK>/<-īK> <-iK> <-iC> <-eC>/<-iC>
Literary Mon <-eK> <-iK> <-eK>

I prefer Diffloth's solution to my own not only because of its firm comparative grounding but also because

- it made use of existing contrasts (short vs. long, i vs. e) instead of eliminating length and introducing a third height (artifacts of the Pre-Proto-Monic reconstruction I didn't post).

- the phonetic shifts that I think it requires make more sense (and are much simpler for Nyah Kur though not for Mon):

Nyah Kur:

1. *-K > *-C after *i(i)

2. *ee > ii

Mon:

1. Neutralization of long vowel height before *-K and *C:

*-eeK > *-iiK

*-iiC > *-eeC

2. Neutralization of vowel height before *-C:  *-iC > *-eC (to match *-eeC)

3. Diphthongization of *e(e) to assimilate to a following *-C:

*-e(e)C > *-eiC

There was no Indic vowel symbol for [ei], so *-eiC was written in Old Mon as both <eC> and <iC>.

4. Loss of vowel length: *-iiK*-iK

I predict that the first attestation of <-īK> with a long vowel predates the first attestation of <-iK>. If that is not the case, then vowel length was already being lost at the time of the earliest known Mon texts, and <-īK> with a long vowel represented a conservative, waning pronunciation.

5. Backing of *-C to *-K after front vowels (dissimilation?)

*-(e)iC > *-(e)iK

6. Straightening of *ei to <e> before *-K (the *-i- was motivated by a following palatal that no longer exists)

*-eiK > <eK>

One could regard 5 and 6 as the same change on an abstract phonemic level:

*/eC/ > /eK/

since */e/ was *[ei] before */C/. (There was no contrast between *-eC and *-eiC.)

Note that some of the above changes are mine and not Diffloth's, though their starting point is his.

I'm embarrassed by how wrong I was, but I'm keeping my original solution in this post anyway as an example of what not to do: namely, fail to exploit existing resources. One approach to solving problems like this one is to calculate the maximum number of possibilities given existing variables and then see which possibilities make for the simplest 'story':

2 heights (i, e) * 2 lengths (short, long) * 2 coda classes (-C, -K) = 8

1. *-iC

2. *-iiC

3. *-iK

4. *-iiK

5. *-eC

6. *-eeC

7. *-eK

8. *-eeK

Although "2 lengths" may seem redundant, there are languages with three vowel lengths (more examples here). (Mon, however, is not one of them.)

*-eC and *-eK can be ruled out at the proto-Monic level because Nyah Kur, the only Monic language with vowel length, lacks short e.

Another error I made was investing too much in wrong hypotheses: the absence of vowel length and the presence of a third front vowel height. If I had worked from the eight possibilities above, I might have seen how a more Nyah Kur-style solution was superior.

Lastly, the improbability of Pre-Nyah Kur *-iiK not becoming *-iiC given *-iK-iC in my reconstruction should have been a warning sign. It would make phonetic sense for Pre Nyah Kur *-iiK to become *-iɨK (i.e., have its second half lose its palatality to assimilate to the following velar) which would then be immune to the change *-iK > *-iC. Then *-iɨK would simplify into to Nyah Kur *-iiK. However, such a complex intermediate stage is obviously an attempt to salvage a flawed reconstruction; it adds unnecessary complexity. Diffloth's solution is much simpler: *-eeK-iiK.


15.11.29.23:52: PROTO-MONIC VOWELS IN DIFFLOTH (1984)

(Posted after expansion on 15.12.17-18.)

I have been reading Gérald Diffloth's The Dvaravati Old Mon Language and Nyah Kur (1984) to understand Monic history and to get a better grasp of vowel warping which also occurred in Chinese and possibly also Tangut. The complex diphthongs,of modern Monic languages come from much simpler Proto-Monic vowels: e.g.,

Singu ʔəsʌe̲i̯a < Proto-Monic *k[r]sw 'to whisper'

The underline and subscript diaresis indicate the most prominent vowel with clear (underline) or breathy (diaresis) phonation. Proto-Monic did not have phonemic phonations (though perhaps the two types of phonation already existed on a subphonemic level). Conversely, the subscript inverted breve indicates the least prominent vowel.

I have not found a diagram of Proto-Monic vowels (as opposed to rhymes) in Diffloth's book, so I made one myself:

*i/*ii length only contrastive before final palatals and velars *ɯ/*ɯɯ only two instances of *ɯɯ known *u/*uu *uu nearly in complementary distribution with *oo
*iə nearly in complementary distribution with *ɛɛ (no *ɯə!) (*uə as alternative to *ɔɔ)
(no *e or *ee!) /*əə only two instances of *əə known -/*oo nearly in complementary distribution with *uu
-/*ɛɛ nearly in complementary distribution with *iə *a/*aa /*ɔɔ length only contrastive before final velars could *ɔɔ be *uə?

The five most common vowels according to Diffloth (1984: 284) are in large type.

Although *ɯ(ɯ) is a back vowel, I have followed Diffloth in grouping it with the central vowels.

I moved the low vowels *a/*aa into the lower mid row for a more compact chart.

If I ignore marginal contrasts, I could reconstruct a more symmetrical Pre-Proto-Monic system with a phonemic length distinction for only one vowel:

*i *u
*iə > *ɛɛ (no *ɯə!) *uə > *ɔɔ
*e > *ii  *o > *oo and *uu
(no *ɛ) *a/*aa

11.30.0:10: I could even eliminate that one last remaining length distinction by positing a chain shift: *ɯə > > *a > *aa.

15.12.17.20:00: But I would rather reconstruct a high-frequency vowel as *aa instead of *ɯə.

Here's how I think front vowels could have developed between Pre-Proto-Monic and Proto-Monic:

Stage 1 Stage 2 Stage 3
*i *i *i/*ii (< *ei < *e)
*iə *iə *iə
*e *e (no *e)
(no *ɛ) *ɛɛ (from *eɛ < *ie < *iə before glottal stops) *ɛɛ

For comparison, here's a table of the development of back vowels:

Stage 1 Stage 2 Stage 3
*u *u *u/*uu (< *ou <  *o before nongrave consonants)
*uə (no *uə) (no *uə)
*o *o *oo (< *ou < *o before grave consonants)
*ɔ/*ɔɔ (from *oɔ < *uɔ < *uə)  *ɔ/*ɔɔ

In both systems, diphthongs (*iə, *uə) monophthongized into long vowels, and mid vowels (*e, *o) became yet more long vowels.

It is not surprising that there is only a three-way contrast between front vowels since front vowels are less frequent than central or back vowels in Proto-Monic. Do other Austroasiatic languages have relatively low frequencies of front vowels? The imbalance of front and back vowels in Pre-Proto-Monic may go back to Proto-Austroasiatic. Shorto's 2006 reconstruction of Proto-Austroasiatic has no corresponding to *ɔ.

12.18.7:28: I am not happy with the above proposals since they require unmotivated splits:

- *iə*iə except

- before glottal stops: *iə*ɛɛ

- in two problematic words without glottal stops (Diffloth 1984: 282-283):

*t[l]m[ɛɛ/aa]t 'flattened' (Nyah Kur points to *ɛɛ, but Mon points to *aa)

*k[ ]ʔɛɛm 'to clear one's throat' (a sound-symbolic exception to sound change?) 

- *o > *oo or *uu before glottal stops (the height is otherwise predictable; see Diffloth 1984: 276-278, 380-381)

One might be able to write off the two unexpected cases of *ɛɛ, but there are ten instances of both *oo and *uu before glottal stops (Diffloth 1984: 377-378). Twenty words are too many to be ignored. If Pre-Proto-Monic had length distinctions for *a and *u (and other central vowels?), perhaps it had one for *i as well:

*i/ii *ɯ/ɯɯ? *u/uu
(no *e/*ee) *ə/əə? (no *o)/*oo
(no *ɛ)/*ɛɛ *a/*aa *ɔ/*ɔɔ

That inventory is almost identical to Diffloth's. We have come almost full circle.

The distribution of distinctive long vowels is identical to that of Nyah Kur (Diffloth 1984: 52).


Tangut fonts by Mojikyo.org Tangut radical and Khitan fonts by Andrew West Jurchen font by Jason Glavy All other content copyright © 2002-2015 AmritavIision