bug: fix the Karen letter from ာ် ာ to ၥ် ၥ on Sgaw Karen Keyboards

[Nnyny]

From the community site (https://community.software.sil.org/t/could-you-please-fix-the-karen-letter-from-to-on-sgaw-karen-keyboards/10527)

Hi,

Recently, the Karen Organisation in Myanmar has modified the correct spelling of one Karen word in all Unicode keyboards. I would like to suggest that you could change that word on Keyman Karen keyboards.
Previously, the Keyman Karen Keyboard displayed the word as ာ် or ာ, whereas the correct spelling is ၥ် or ၥ, without any slant.
This keyboard was created by the Main Karen Organisation in Myanmar, and they correctly displayed the correct spelling of these words.

Kawthoolei Unicode

Kawthoolei Unicode
Modernizing the Karen language: Unicode‑standard fonts, keyboards, and tools by the KNU’s Karen Education & Culture Department and KECS.

The keyboards to fix in Keyman:

Sgaw Karen keyboard
sgaw_karen keyboard
knyaw keyboard
Thank you so much for your help.

[LornaSIL]

Is this a keyboard issue, or is it possibly a lexical model that needs updating?

[DavidLRowe]

I'm not finding a lexical model in the Keyman lexical-models repository for "ksw".

[DavidLRowe]

knyaw keyboard:
[K_M] > U+102C U+103A

sgaw_karen keyboard:
[K_3] > U+102C U+103A
(2 other rules that output U+102C with other characters)

sil_sgaw_karen keyboard:
[K_M] > U+102C U+103A
[SHIFT K_M] > U+102C
(1 other rule that outputs U+102C with other characters)

I don't see any combination in any of the 3 keyboards that outputs U+1065

I'm not exactly clear on what the desired change is.

[hexadecane]

You could find the word difference in this website and keyboard. https://unicode.kawthoolei.app

I am not sure which unicode number do they use but the word ာ် and ာ in Sgaw karen keyboards are not correct since that spelling will create the word like this အာ် but instead it should be အၥ် without connecting to each other. The correct spelling is ၥ် and ၥ.

Maybe you could find those characters in Burmese or Myanmar Unicode chart.
And also it could be combination of ၥ and ် from the myanmar unicode chart.

Thanks for your helping.

[hexadecane]

Is this a keyboard issue, or is it possibly a lexical model that needs updating?

It is the issue of the characters on the keyboard. Not the keyboard issue. Thanks.

[andjc]

It is more than just an issue related to keyboards, it has implications for a wide range of things including sorting, line breaking, etc.

[DavidLRowe]

The characters we're discussing are:

Unicode value	Character	Unicode name
U+102C	ာ	MYANMAR VOWEL SIGN AA
U+103A	်	MYANMAR SIGN ASAT
U+1065	ၥ	MYANMAR LETTER WESTERN PWO KAREN THA

I can see that U+1065 and U+102C are visually similar, though the second is a vowel that attaches to a preceding character.

It seems unusual, however, to remove U+102C from the keyboard. Isn't it the case that users might need to type that vowel in their text? And they might need to type the U+102C U+103A combination and benefit from having that on one key.

I'm not familiar with the S'gaw Karen language or the Myanmar script. Looking at the English Wikipedia article on the S'gaw Karen language, I see that U+102C occurs 5 times, U+103A occurs 31 times and U+1065 doesn't occur at all in that article. I'm not claiming that Wikipedia is an authority, but I'd be reluctant to remove characters from a keyboard when they are used in an article describing the language the keyboard was designed to support.

Can we simply add the U+1065 to the keyboard as another character that can be typed? If so, where should it be placed on the keyboard?

Thanks for your patience as we work through this issue.

[hexadecane]

Thanks for the reply. I know you will see the spelling combination of U+102C U+103A since that spelling is taken from Burmese language characters and used in so many texts only since they cannot provide proper Sgaw karen characters. The actual and correct combination is U+1065 U+103A, and the word should come out like this without any slant အၥ်.

If you can replace ၥ် in the place of ာ် in all three sgaw karen keyboards in Keyman that i have mentioned earlier.

Besides if you could add မ် and ဒ် in where space available since those words should be combined together without spelling it individually.

I will write down all the Sgaw Karen only characters and if you could remove any characters that is not included here. Because people are mixing up with Sgaw Karen keyboard and Burmese Keyboards.

Sgaw Karen Correct Characters.

က ခ ဂ ဃ င
စ ဆ ၡ ည တ
ထ ဒ န ပ ဖ
ဘ မ ယ ရ လ
ဝ သ ဟ အ ဧ

ှ ၠ ြ ျ ွ မ် ဒ်

ါ ံ ု ူ ့ ဲ ိ ီ ၢ ် ၢ် ၥ် း ၣ် ၤ

I can see that in Sgaw Karen keyboard they have added unusable characters such as

ဗ in A, ဇ in Z, ာ in M, ဉ in N, ေ in a

if you can remove those characters and replace this two words မ် ဒ် anywhere there is space available.

You can check the correct characters in Karen Organisation website

https://unicode.kawthoolei.app

They have a screenshot of a keyboard and characters. Thanks for your help.

[mcdurdin]

tl;dr: Can we engage both @mhosken and @sdysart on this? I don't have any specific expertise in these languages, and I think it is important that we engage with those who do.

U+102C was assigned in v3.0 in 1999 when Myanmar script was first encoded. U+1065 was assigned in v5.1 in 2008 for Western Pwo Karen.

Martin Hosken wrote UTN-11 Representing Myanmar in Unicode that somewhat backs up the request in this issue, both in its S'gaw Karen description, and in the discussion of U+102C (highlights mine; copied from PDF so no guarantees on codepoints here):

In this example, we highlight the code of interest. Notice how the ါ (U+102B MYANMAR VOWEL SIGN TALL AA) has a different code to the ာ (U+102C MYANMAR VOWEL SIGN AA). The Myanmar character underlying the two codes is the same, and there are rendering rules that can give the correct form, so why has the tall -aa been given its own code? The primary reason is that Sgaw Karen, among other minority scripts, only has the tall form, and so a rendering system that works for the Myanmar language is not going to work for Sgaw Karen and vice versa. A Myanmar language specific keyboarding implementation could choose to enforce a particular variant of the -aa vowel in the context of certain consonants (in Burmese following ခ, ဂ, င, ဒ, ဓ 2 , ပ, or ဝ), medial combinations and syllable chainings, but this is not required.

However, the Sgaw Karen (ksw) section mentions neither 102C nor 1065.

Other Karennic language usage of these letters, according to UTN-11:

• Western Pwo Karen (pwo) - 1065
• Pa'o Karen (blk) - 102C
• Eastern Pwo Karen (kjp) - 102C
• Kayah (kyu) (Myanmar script) - neither

All of the referenced Keyman keyboards support only Sgaw Karen (ksw) according to their language metadata:

• sil_sgaw_karen, created by @sdysart in 2017, maintained by @MakaraSok
• sgaw_karen, created by @Jcturner12 in [sgaw_karen] Initial submission of sgaw_karen keyboard #2752
• knyaw, created by @Ehblut in [kynaw] initial submission of knyaw keyboard #2575

sil_sgaw_karen specifically notes that it includes letters from Burmese, including 102C:

Since the majority of users of S'gaw Karen keyboard will likely be bilingual with Burmese, some Burmese characters are added to the keyboard for convenience purposes. Users do not have to switch keyboards to type characters from loanwords. Those characters are: ်, ာ, ​ေ, ဗ, ဉ and ယ.

The other two keyboards have no narrative on Burmese letter inclusion.

I think we should go back to the keyboard authors and get their perspective.

(A secondary issue is we have these three Sgaw Karen keyboards with absolutely nothing in the descriptions to distinguish between them; even the Author is blank for two of the keyboards. This should probably be improved!)

[andjc]

A broader consultation would be a good idea, the change of U+102C to U+1065 will have wider implications, not just to keyboards. It would impact line breaking, collation/sorting, syllabification and tokenisation, along with other things. Also some of the character properties of U+1065 will be changing in Unicode 17.0 so there may be differences between applications that use character data from different versions of Unicode.

…

On Thu, 21 Aug 2025 at 13:52, Marc Durdin ***@***.***> wrote: *mcdurdin* left a comment (keymanapp/keyboards#3673) <#3673 (comment)> tl;dr: Can we engage both @mhosken <https://github.com/mhosken> and @sdysart <https://github.com/sdysart> on this? I don't have any specific expertise in these languages, and I think it is important that we engage with those who do. U+102C was assigned in v3.0 in 1999 when Myanmar script was first encoded. U+1065 was assigned in v5.1 in 2008 for Western Pwo Karen. Martin Hosken wrote UTN-11 Representing Myanmar in Unicode <https://www.unicode.org/notes/tn11/UTN11_4.pdf> that somewhat backs up the request in this issue, both in its S'gaw Karen description, and in the discussion of U+102C (highlights mine): In this example, we highlight the code of interest. Notice how the ါါ (U+102B MYANMAR VOWEL SIGN TALL AA) has a different code to the ါာ (U+102C MYANMAR VOWEL SIGN AA). The Myanmar character underlying the two codes is the same, and there are rendering rules that can give the correct form, so why has the tall -aa been given its own code? *The primary reason is that Sgaw Karen, among other minority scripts, only has the tall form,* and so a rendering system that works for the Myanmar language is not going to work for Sgaw Karen and vice versa. A Myanmar language specific keyboarding implementation could choose to enforce a particular variant of the -aa vowel in the context of certain consonants (in Burmese following ခ, ဂ, င, ဒ, ဓ 2 , ပ, or ဝ), medial combinations and syllable chainings, but this is not required. *However, the Sgaw Karen (ksw) section mentions neither 102C nor 1065.* Other Karennic language usage of these letters, according to UTN-11: - Western Pwo Karen (pwo) - 1065 - Pa'o Karen (blk) - 102C - Eastern Pwo Karen (kjp) - 102C - Kayah (kyu) (Myanmar script) - neither All of the referenced Keyman keyboards support only Sgaw Karen (ksw) according to their language metadata: - sil_sgaw_karen, created by @sdysart <https://github.com/sdysart> in 2017, maintained by @MakaraSok <https://github.com/MakaraSok> - sgaw_karen, created by @Jcturner12 <https://github.com/Jcturner12> in #2752 <#2752> - knyaw, created by @Ehblut <https://github.com/Ehblut> in #2575 <#2575> sil_sgaw_karen specifically notes <https://github.com/keymanapp/keyboards/blob/ab3bb647e63099d70c0434bc999974350ce08577/release/sil/sil_sgaw_karen/source/welcome/welcome.htm#L21> that it includes letters from Burmese, including 102C: Since the majority of users of S'gaw Karen keyboard will likely be bilingual with Burmese, some Burmese characters are added to the keyboard for convenience purposes. Users do not have to switch keyboards to type characters from loanwords. Those characters are: ်, ာ, ​ေ, ဗ, ဉ and ယ. The other two keyboards have no narrative on Burmese letter inclusion. I think we should go back to the keyboard authors and get their perspective. (A secondary issue is we have these three Sgaw Karen keyboards with absolutely nothing in the descriptions to distinguish between them; even the Author is blank for two of the keyboards. This should probably be improved!) — Reply to this email directly, view it on GitHub <#3673 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AALGM67GZUZRC3TC6ETCQUL3OU7BLAVCNFSM6AAAAACD34XJFKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTEMBYHEYDENRTGU> . You are receiving this because you commented.Message ID: ***@***.***>

-- Andrew Cunningham ***@***.***

[mhosken]

Switching from U+102C to U+1065 is huge and it is often far too easy to get backing from part of a government. We don't know if this is going to fly or not? This is especially difficult given the current situation. I can see why they've done it, because ta has a more upright look to it and that fits with the style.

What are some of the technical implications of making the switch:

• Since the sequence is always going to U+1065 U+103A, then the ta is being killed and should remain part of the previous syllable and so they can get away with this because they don't have 102C occurring without 103A. If they did, then there would be problems.

• It will introduce a cursor position before the tone mark that was not there before

• It will mess up sorting and we'll need to adapt the sort order for Karen

So not much technically. The biggest impact will be a bifurcation of the data set for Sgaw Karen causing all kinds of confusion over many years, which is sad. First they have to get everyone to agree to the change and then they have to recode the data. They would be better off just asking for a better styled 102C in that combination which would not be hard to deliver. It's a pretty easy contextual substitution in OpenType (102C > 1065) 103A.

@andjc what are you seeing coming down the pipe that might impact this?

[andjc]

@mhosken, the Indic_Conjunct_Break (InCB) property value for U+1065 is None, but in 17.0 that will be changing to Consonant. I'd need to do some testing, but that would have to wait until 17.0 support is implemented. I have some web tests, but that will need to wait till 17.0 support rolls out.

I would expect that layout engines would break before the tone mark.

[hexadecane]

Is there any updates regarding this issue? Thanks

[mcdurdin]

@hexadecane I understand where you are coming from. It's difficult to know where to go with this, because we haven't heard from any perspective from the other keyboard authors @Jcturner12 and @Ehblut. Again, I am not an expert on this writing system so am dependent on others. @andjc and @mhosken have given their perspectives, but so far I am not seeing a consensus on the change.