Re: FVS Assignment Mismatch WrapUp - GA

On Sun, 23 Aug 2015 15:19:51 +0000
Greg Eck <> wrote:

> I am ready to wrap up the discussion on FVS Assignment Mismatch.
> However I am still lacking good examples on two of the over-rides
> discussed ...
> *         182D Medial - given the case where the contextual rules for
> the dual dots must be over-ridden. In other words, the context
> dictates that the medial GA is dotted, however, the actual shaping of
> the word is desired without the dots. I have not had the time to
> track down examples for this.
> *         182D Final - given the case where the feminine final GA
> does not follow the common pattern of sweeping to the left, but
> however sweeps to the right. In other words, the word is composed of
> feminine vowels, but carries a masculine right-ward swept tail. From
> discussions with Professor Quejngzhabu, I understand that there are
> just a small subset of words (5-6 in quantity) that follow this
> pattern.

You have now found examples of 182D final while I was waiting for my
copy of the Chinese standard (GB/T 26226-2010).  I think the best
example to quote is the one in Row 15 (sig) of Table 9 of GB/T
26226-2010 = Row 17 of table straddling pp5-6 of TR 170.  
(As a matter of curiosity, for I suspect it is irrelevant for rendering
suffixes, are the words with the masculine final actually feminine?)

Both documents show that final <U+182D, U+180B> is context sensitive -
it has the opposite gender to final U+182D. There is therefore no need
to add a variation selection for unexpected final masculine, such as
Professor Quejingzhabu's 2000 document's <U+182D, U+180B>.  The toggle
behaviour is consistent with the graphics in the Unicode charts - Ken
Whistler has already indicated that the UCD and code charts do not need
to show that a variation selector acts as a toggle.

On Mon, 24 Aug 2015 13:58:54 +0000
Greg Eck <> wrote:

> I am still not sure that we have a case for the Medial GA undotted
> over-ride FVS. More thoughts here ... ?

Initial and medial <U+182D, U+180B> are also context-sensitive, having
the opposite dot setting to U+182D.  I have probably oversimplified
the rules expressed in the attachments to , worked
on by Martin Heijdra and Timothy Partridge.  Unfortunately, that work
seems to use slightly different variation sequences to the ISO/Unicode
and Chinese standards.

As I said before, work on variant forms for connected text cannot be
done properly without identification of the contextual shaping rules.

It is also a bad idea to try to include grammar rules in rendering
rules.  For example, some dialects have the rule that /e/ in
non-initial syllables does not overrule an earlier masculine vowel when
it comes to the gender for the suffixes, and it seems that that also
applies to final consonants.  It seems much simpler, and flexible, to
ignore that rule.

There was some horrifying text in TR170:

"The mechanism of inputting characters is not specified by the standard,
so any keyboard driver capable of generating the appropriate 16-bit
character encodings can be used.  However, the input mechanism should
ideally generate the correct positional forms, variants and ligatures
on input by analysis of the context of each letter, at least where

This suggests that the contextual rules could vary from input system
to input system, with decisions on rendering being stored by the use
of the PUA.  This is obviously not consistent with the idea of storing
Mongolian text just using assigned Unicode codepoints.  


Received on Wednesday, 26 August 2015 07:11:11 UTC