Breton collation

[Pierrick Brihaye]

Hello,

I hope I am on the right list...

This is my point. It is about breton language and its specific characters, that is 

Ꝃ ( U+A742) and  ꝃ ( U+A743 ), which are respectively contactions of "Ker" and "ker" at the beginning of family names (my mother's name is "Kergresse") and place names (I live near "Kervos"). It was widely used.

"ker" means "a populated place", wether a house or a megalopolis (yes !).

Regarding the character, see :  https://fr.wikipedia.org/wiki/%EA%9D%82

This character was widely used in breton but french law made it disappear :
1881 : order form the Navy (there were and there are still many bretons sailors)
1895 : prohibition in official use for administrative acts (State Council)
1955 : official and definitive disparition since the  k with diagonal stroke is a genuine orthographic alteration (sic ! )

It is still used in genealogy or historic scientific articles...

My question is about the way it may be sorted. I'd like to add something about the expansion towards Ker/ker in breton AS WELL AS in french (because french was used for civil registration since the 16th century at least to 1895).

Regarding alphabetic order, we have  Keq < Ꝃ < Ker  (no worry : there is no Q in breton)
and   keq <  ꝃ   < ker

Can you tell me to whom I should ask for official implementations ?

Thank you,

p.b.

[Markus Scherer]

Hi Pierrick,

On Tue, Feb 3, 2026 at 5:07 AM 'Pierrick Brihaye' via icu-design <icu-d...@unicode.org> wrote:

I hope I am on the right list...

Not quite.

https://icu.unicode.org/contacts

icu-design --> “This list is for API proposals and their discussion.”

It also depends a little on whether you are asking for help for how to do something (icu-support list), or whether you are aiming to change the locale data.

My question is about the way it may be sorted. I'd like to add something about the expansion towards Ker/ker in breton AS WELL AS in french (because french was used for civil registration since the 16th century at least to 1895).

Regarding alphabetic order, we have  Keq < Ꝃ < Ker  (no worry : there is no Q in breton)
and   keq <  ꝃ   < ker

First off, the Breton collation tailoring (the rules that make it different from the Unicode default sort order) are here:

https://github.com/unicode-org/cldr/blob/main/common/collation/br.xml

You didn't say what level of difference you expect between ꝃ and ker.

See https://cldr.unicode.org/index/cldr-spec/collation-guidelines

Usually, a ligature or an additional diacritic adds a secondary-level difference.

To get what you are asking for, you could add a collation tailoring rule like

&[before 2]ker<<ꝃ<<<Ꝃ

which sorts like this:

<1 keq
<3 Keq
<1 ꝃ
<3 Ꝃ
<2 ker
<3 Ker

where <1 is a primary-level difference (letter-like), <2 secondary (ligatures, diacritics), <3 tertiary (case-like).

Are you sure that you want ꝃ < ker ?

I would expect the letter-with-stroke to sort after the sequence.

That would be

&ker<<ꝃ<<<Ꝃ

-->

<1 keq
<3 Keq
<1 ker
<3 Ker
<2 ꝃ
<3 Ꝃ

I used the ICU Collation Demo for this, editing the “Append rules” and populating the Input box.

Do you have a reference for how this should sort?

Or examples from phonebooks, dictionaries, or other sorted lists which can help distinguish between orders and levels?

Can you tell me to whom I should ask for official implementations ?

Eventually, probably a CLDR ticket (link above).

Note that the French sort order is currently the same as the Unicode default sort order. Adding a tailoring has a performance impact, and doing so for a historical character will face hurdles.

It's probably easier to argue for its support in the Breton collation, and recommending to use that for French if necessary.

Best regards,

markus

[pierrick...@free.fr]

Hello,

First, thank you for the link to the breton collation that considers the trigraph "c'h" which, indeed, has a special treatment in the collation of breton language.

In fact, I don't know if I want ꝃ < ker or ker < ꝃ because I don't know of any document with the 2 forms coexisting !

I've built such a proposition because I consider that "k with diagonal stroke" (https://www.compart.com/fr/unicode/U+A742) should imply that the stroke as a carried by the letter "k" and that "k" is before "ker"... I admit that this could be discussed.

Regarding level of differences, I would think that it is total. "abc" is different from "a", isn't it ? But maybe you can provide me an exemple of such a significant difference treatment ?

For string equality, ꝃ is strictly equivalent to "ker".

So, what document should I propose to make an official request ?

Thank you anyway for your help.


Cheers,


p.b.



----- Mail original -----
De: "Markus Scherer" <marku...@gmail.com>
À: "Pierrick Brihaye" <pierrick...@free.fr>
Cc: "icu-design" <icu-d...@unicode.org>
Envoyé: Mardi 3 Février 2026 20:35:37
Objet: Re: Breton collation

[Markus Scherer]

On Tue, Feb 3, 2026 at 1:18 PM <pierrick...@free.fr> wrote:

In fact, I don't know if I want ꝃ < ker or ker < ꝃ because I don't know of any document with the 2 forms coexisting !

Then I would go with what is mostly done, and treat the variant as sorting after the base form.

Regarding level of differences, I would think that it is total. "abc" is different from "a", isn't it ? But maybe you can provide me an exemple of such a significant difference treatment ?

For string equality, ꝃ is strictly equivalent to "ker".

This means that you want ꝃ=ker on a primary (letter-like) level.

Unless there is better information, a secondary (diacritic-like) difference should be appropriate.

This would be like in the German "phonebook" sort order:  &AE<<ä<<<Ä  etc.

Again, see https://cldr.unicode.org/index/cldr-spec/collation-guidelines for details.

So, what document should I propose to make an official request ?

Please do look for an authoritative reference.

Given that this is not a modern-use character, you probably need to look for examples from phonebooks, dictionaries, lists of book titles, or other sorted lists which can help distinguish between orders and levels.

Once you have one or more good references, you can create a CLDR ticket:

https://cldr.unicode.org/requesting_changes

Best regards,

markus