Some remaining issues, mainly Manchu

Hi all:

I have been very happy to see you all working out the many issues with Mongolian, and did not think it was my place to interfere with any possible solution.

However, things have moved on to Todo, Sibe and Manchu. When I prepared the Baiti font for Microsoft (not 100% the font which finally shipped), especially for Manchu I had to add undocumented FVS1 and FVS2 to cover exceptions not dealt with in the documentation I had received from Quejingzhabu. Since then,  I have come across some other oddities. I now list them here, not to propose particular solutions, but just to make sure you and future developers know that the cases need to be dealt with one way or the other. Thus, the following examples I found the most difficult to deal with, ( I added FVSs just unilaterally and temporarily in the font, but you may not have done so without proper documentation.)

For Manchu, I once prepared a handout for users of Manchu, based upon the original versions of the Baiti font. I include that document here, since it does include cases to be covered (the keyboard refers to a keyboard I made for some users.)


1.       My first question is actually about a Mongolian word I had difficulty with: how do you propose users input the Mongolian word for the Chinese guoshi; dictionaries vary in how they read/spell the word, guosi, goosi or güüsi, but the spelling/input is not obvious. In Western books introducing Classical Mongolian, it is usually  spelled guosi, introduced from the very beginning, but it cannot be derived at that way using the current fonts.


2.       Then for Manchu, some necessary issues to deal with:



·         Also for Manchu, the question is, what is the proper default for the -i in diphthongs. All textbooks, and most dictionaries, will tell you the double oblique variant is the default. One major Chinese dictionary however has the straight-plus-oblique variant as default. Quejingzhabu followed that dictionary; that solution is strongly resisted by any actual Manchu user. Users will need to be able to arrive at either; and possibly a FVS2 needs to be invoked. This needs to be discussed. See 1c, 2i, 3a in my Manchu document.


·         Please realize that the Sibe syllable marker needs to be implemented for Manchu; otherwise many names and  loanwords from Chinese  cannot be spelled. See e.g. rule L in Manchu document.


·         There is a couple of well-known exceptions which need to be dealt with involving Manchu k- and h-: welhūme, nehū, nehūji, bukūn, bukūri, kūke. See 2d In Manchu document.  I added FVS1.


·         The weirdest example is a rare word kuiiii, spoon (also spelled kuii—the issue is, can it be derived at any which way). It has 3 obliques usually for those i’s (plus final -i); but perhaps, if following the dictionary which has stick-stroke form, it should also be able to be spelled straight-oblique-straight (plus final). The word is only found in dictionaries so far; its further status in uncertain, thus any solution is OK. By using the Sibe syllable marker (needed anyway) and using a possible –y- as input, I managed to limit somewhat the need for multiple FVSs. But both versions need to be derivable. See rule L.



3.       Finally, I have found three sets of glyphs not present in any font I know. The first is the known Mongolian swatches case (see document),which I think you know but decided it was not a current issue,  while I found the other two cases after I did my work, so I did not hide solutions in the font (as I did with the Ali Gali for LH, which is now added as its own character)


·         Todo Ali Gali. Shagdarsürüng and some other fonts give an alternative version of Todo Ali Gali long vowels ī, ū (and syllabic long r and l: ṝ, ḹ). They involve versions of 1847 u plus long vowel marker 1843; the Mongolian (not Todo!) final 1822 plus Todo long vowel marker (yes! Not Todo –i+long vowel marker); plus a recombined long vowel marker + final Mlta. half ya (18A7) ligature after r- and l-for the long syllabic r and l (I assume logical input would be 18A7+1843, as for ā). I have only seen isolated Ali Gali syllables; I assume relevant initials and medials are also in order. Currently, there are no glyphs for  1847+1843, 1822+1843, and 18A7+1843 (visually, 1843+18A7).


·         The strangest discovery: there are actually two versions of the Tong wen yun tong, basis of Manchu Ali Gali. People usually only know of the shorter one, with all table structures for transliterations  from Sanskrit and, separately, Tibetan. That one has been republished frequently. There is however another versions with the same tables, plus with each possible syllable actually spelled out, republished in 1978 in Taiwan. Going in detail through the latter, there are two issues never really remarked upon:


(1)    the middle “e” in syllables ending in –eng is often left out (tngri is well-known, but it happens more often), a fact with no influence whatsoever for fonts and input, but also:


(2) If certain consonants with circles are followed by a vowel with a dot (e, ue), the diacritics change place. This does not happen with all consonants with circles, but if it happens, it always does: that is, when it happens with –ue, it will happen with long –ue, -e and long –e: it is not an accidental miswriting. I have never seen a font doing this, nor any textbook pointing this out. It probably is best treated as optional variants, just as “bl” of “gl” ligatures; but very properly, this should be possible in a font. The three cases are:

                                                               i.      MLAG JHA           189D + ue, e : circle and dot switch places (see here the twyt001 document, p. 90)

                                                             ii.      MLAG DDHA:     189F + ue, e : circle-circle-dot become circle-dot-circle (p92)

                                                            iii.      MLAG BHA:        18A8 + ue, e: circle+ dot become dot-circle (p 97), resulting in new ligatures

Hope someone will take notice…

Martin Heijdra