15.10.17.23:59: HOW DO WE WRITE PHONETICALLY OBSCURE PROTO-PHONEMIC DISTINCTIONS? THE CASE OF RATLIFF'S PROTO-HMONG-MIEN

Last night I finally got ahold of Martha Ratliff's 2010 book Hmong-Mien Language History. I had been looking forward to it for years because what little I had seen of her reconstruction looked much more plausible than Wang Fushi and Mao Zongwu's from 1995. I've read up to page 33 so far, and have been impressed by her honesty: e.g.,

Yet there are remaining problems. First the over-reconstruction of rimes is still troublesome: it will be important to find conditioning factors to reduce the Proto Hmong-Mien rime inventory. Wang and Mao's reconstruction of every possible low vowel on the IPA chart (*æ, *a, *A, *ɐ, *ɑ, *ɒ) appears to be a clear example of "over-reconstruction", but one could also say that this reconstruction is just as unnatural: what present-day language makes a distinction between /ei/ and /ej/, between /æi/ and /æj/, or worse, among /ɔu/, /ou/, and /ow/ - rimes claimed here to be distinctive in Proto Hmong-Mien? It is clear that these values are abstractions, and only capture the true nature of the protolanguage in a shadowy fashion. (pp. 29-30)

The monosyllabic rhymes  /Vi/ and /Vj/ look like alternate notations for the same segmental sequence. (If /Vi/ were disyllabic /V.i/, it would certainly be distinct from monosyllabic /Vj/.)

I don't think a distinction between /ɔu/ and /ou/ is impossible, but /ou/ and /ow/ have the same problem as /Vi/ and /Vj/.

In Tangut reconstruction, I have encountered some other approaches to this problem:

- Gong simply reconstructed pairs of rhymes as identical: e.g., 36 and 37 as -jij.

The most troublesome approach because it does not indicate distinctions unless rhyme numbers are listed.

- Arakawa (2014) used numbers to distinguish between similar rhymes: e.g., 94 as -yer and 101 as -yer2.

The number might be mistaken for a tone number (though Arakawa writes tone numbers at the beginning of syllables), and even if it isn't, it may imply that all the -2 rhymes are similar in some way (other than being phonetically similar to more common rhymes written without -2)

- Arakawa and I use -' (which I have called 'prime' though it is not a prime symbol) which I can combine with grade numbers to distinguish between similar rhymes: e.g., 34-40 in my notation are -e1, -e2, -e3, -e4, -e'1, -e'2, -e'3 ~ -e'4.

The prime symbol may be misinterpreted as a glottal stop, and grade numbers are not an option in a language without them.

I am not sure how I would handle the problem of distinguishing between similar glide-final rhymes in Proto-Hmong-Mien.

10.18.13:12: Although I would be initially tempted to rewrite the less common glide/vowel as /j2 w2/ or /j' w'/, there is no correlation between final glide/vowel notation and frequency in Ratliff's reconstruction: e.g., /(u̯)ei/ is more common than /ej/, but there is an /aj/ without an /ai/, so one cannot say that /i/ is more common than /j/. Numbers of reconstructed words per rhymes are listed below with Chinese loans in parentheses: e.g., /əi/ is only in /Kəi/ 'chicken' (an areal word if not a borrowing from Chinese 雞). Bold indicates the more common (or sole) member of a pair: e.g., /uj/ is more common than /ui/.

There is also no correlation between final glide/vowel notation and the French-like dissimilation in Hmongic described in Ratliff (2010: 24):

- dissimilation occurred in /ɛj əj aj/ and /ei æi/:

Proto-Hmong-Mien /ɛj/ > Proto-Hmongic *u̯ɛ

Proto-Hmong-Mien /ei æi/ > Proto-Hmongic *u̯ei

Proto-Hmong-Mien /əj aj/ > Proto-Hmongic *u̯a

cf. Old French -ei > Modern French -oi [wa]

- dissimilation did not occur in /ej æj/ and /əi/:

Proto-Hmong-Mien /ej əi/ > Proto-Hmongic *e (see p. 16)

Proto-Hmong-Mien /æj/ > Proto-Hmongic *æ

Perhaps one could arbitrarily write the dissimilating group with one symbol and write the nondissimilating group with another symbol. But I do not know of any similar criterion for writing the distinction between the /-w/ and /-u/ final rhymes.