Re: Overide #2 (NA and GA)

On Tue, 1 Sep 2015 07:54:28 +0900
<> wrote:

> We are only included medial vowel cases. The the
> word _kino_ will be excluded from our logic. Maybe I can improve the
> expression to match our font current logic.

> O U+1828 N ←First Medial Form (un-dotted medial NA)
> If   (Initial_Vowel+N+Mongolian_Medial Vowel+ is not final FVS1-3)
> OR  (Medial_Vowel+N+Mongolian_Medial Vowel+ is not final FVS1-3) then
> N ← Second Medial Form (dotted medial NA)

I think you're making the logic too complicated.  I would simply write
(though one may want a wordier syntax), if we are assuming NA has
overrides and not toggles,

gN0mu > gN0md / _ cVowel_NA

where cVowel_NA is then defined as some set of vowels.

(gN0mu is the default *g*lyph for *N*A with *0* valid variation selector
for *m*edial position.  I use *u* for undotted, and *d* for dotted.)

I don't believe it matters what precedes the medial NA.

> But as my point of view, it is best to NIRUGU under this medial NA to
> remain un-dotted form. The NIRUGU will prevent the contextual
> override of the following vowel.

"Override" is a confusing word here.  I would just say "change" or
"modification".  "Override" refers to something unusual.

As NIRUGU can be used merely to increase the spacing between the
centres of letters, I think it is wrong to use NIRUGU this way.  The
contextual substitution rules should be unaffected by NIRUGU.  (In OT,
one could declare it a mark in the GDEF table and then restore its
proper advancewidth in the dist feature under GPOS.)

> > What of compounds whose second elements start <NA, OE> or <NA,
> > UE>? Do they use the first or second medial form, and if the
> > UE>latter, is not the NA dotted?  
> > One needs a complete list of vowels. Some may need more research
> > than others - Todo, Sibe and Manchu vowels all need to be
> > considered.

> What I mean here is, for example, the medial OE, UE if it is in the
> first syllable, it will be override to medial <OE, FVS1>, medial <UE,
> FVS1>.

If there are no further contextual changes, I would write this as

gOE0m > gOE1m / cInitial _
gUE0m > gUE1m / cInitial _

(The list would be longer if FVS1 were a toggle.)

I don't even want to handle cases with two consonants before the vowel
- I would require the user or input mechanism to command an override by
storing FVS1.

cInitial is the set of glyphs for letters in initial position.

If we order this change after the selection of the form of NA, then
this case is not a problem for your expression of the rules.

My thought - which may be wrong - is that if a word starting <NA, OE, y>
or <NA, UE, y> is the second element of a compound, its start will (or
may) be stored as <NA, OE, FVS1, y> or <NA, UE, FVS1, y>.  ('y' is
any Mongolian script joining or join-causing character.)  The word's
appearance  would then show that the second element of the compound
word was feminine. Compare 'Hohhot', which switches from feminine to
masculine. I would then expect the NA, which would be medial, as this
is the *second* element of the word, to be dotted.


Received on Tuesday, 1 September 2015 01:30:29 UTC