[Latingp] AE, OE, and Ligatures

Tue Feb 12 22:25:16 UTC 2019

The way I read Dennis' comments above (and what I took from the discussion in Brussels) was that he was arguing that the two ligatures were not variants of each other.  Even though they clearly fit the criteria of "Glyphs nearly identical due to font design"  If I have misunderstood the intent there, my apologies. 

 Bill Jouris
Inside Products
bill.jouris at insidethestack.com
831-659-8360
925-855-9512 (direct)

      From: Meikal Mumin <meikal at mumin.de>
 To: "bill.jouris at insidethestack.com" <bill.jouris at insidethestack.com>; "latingp at icann.org" <latingp at icann.org>; "Tan Tanaka, Dennis" <dtantanaka at verisign.com> 
 Sent: Tuesday, February 12, 2019 6:10 AM
 Subject: Re: [Latingp] AE, OE, and Ligatures

Dear colleagues,
To chime into that debate:
The visual analysis whereby the two glyphs of the ligatures “æ” and “œ” are confusable across a significant number of fonts is language independent, since irrespective of the language of the user or the resources available at an IDN, the rendering of the url may make use of such fonts. The significant number seems independent of the assumed spread of use of single fonts, which we would have to establish in any case (and I doubt we could do that).
Meanwhile any analysis whereby the ligatures are interchangeable with the sequences the same consist of is language dependent. Since we had many similar discussions in ArabGP, let me remind the GP that we are building a ruleset for the root zone, which applies to the entire script-using community, since IDNs do not have a language property. Accordingly, we should not apply language specific rules across languages (unless code-points are restricted in use to few or single language communities, such as German ß - which is clearly not the case with a, o, and e) and therefore it seems irrelevant if e.g. French users consider the consider the ligature “œ” as equivalent to the sequence o+e or English users consider the ligature  “æ” as equivalent to the sequence a+e, unless we can demonstrate the use across a significant part of the script using community, that is across languages.
Hopefully this helps to settle this debate from a methodological point of view.
Best,

MeikalAm 12. Feb. 2019, 14:45 +0100 schrieb Tan Tanaka, Dennis via Latingp <latingp at icann.org>:

On the contrary. Let me explain:    
   - Our “visual” methodology consists in a 4-point scale that assigns a value to the visual similarity of a pair. Each pair of characters is rendered in three different font types. We agreed on Arial, Times and Courier to do such comparison. Each pair is then score independently (e.g. 1 = Identical; 2 = Nearly identical; 3 = Distinguishable; 4 = Different).
   - Our “non-visual” methodology is looking for to prove or disprove the notion that certain handwriting customs are transferred to font design. For this analysis we picked the website wordmark.it to analyze a large number of font types.
 The argument about visual similarity of the ligatures “æ” and “œ” is logically under the “visual” methodology. So, if anything, I’m advocating for us to stick with our methodology and not to apply ad-hoc criteria on a case by case basis. Dennis From: Bill Jouris <bill.jouris at insidethestack.com>
Reply-To: Bill Jouris <bill.jouris at insidethestack.com>
Date: Monday, February 11, 2019 at 4:09 PM
To: Dennis Tan Tanaka <dtantanaka at verisign.com>, "latingp at icann.org" <latingp at icann.org>
Subject: [EXTERNAL] Re: [Latingp] AE, OE, and Ligatures And yet, in our previous work, we considered high visual similarity in a significant number of fonts in workmark to be sufficient cause to consider something a variant. 

So are you arguing for changing our criteria?  Bill Jouris
Inside Products
bill.jouris at insidethestack.com
831-659-8360
925-855-9512 (direct) From: "Tan Tanaka, Dennis via Latingp" <latingp at icann.org>
To: "Michael.Bauland at knipp.de" <Michael.Bauland at knipp.de>; "latingp at icann.org" <latingp at icann.org>
Sent: Monday, February 11, 2019 12:54 PM
Subject: Re: [Latingp] AE, OE, and Ligatures Looking again at the visual evidence, I believe this is not a strong case for variants. But it may be a candidate for visual similarity. From a visual standpoint I don’t see it as a clear-cut case. Doing a comparison of the code points "æ" "œ" using wordmark.it, the great majority of fonts show them very distinguishable (e.g. Arial: æ œ, Times: æ œ, Courier: æ œ, Calibri: æ œ). From an orthography viewpoint, I don’t see good support, but of course, this is only one data point. -Dennis On 2/11/19, 9:16 AM, "Latingp on behalf of Michael Bauland" <latingp-bounces at icann.org on behalf of Michael.Bauland at knipp.de> wrote:     Hi Bill,       thanks for the summary.       I agree with you if purely looking at the visual confusability issue.    However, with ae vs. æ and oe vs. œ the issue is not about visual    sameness, at least that's what the IP argued on our phone call. They    said that in most languages ae and æ have the same meaning and could be    exchanged. I can only talk for German, and there it might be ok to write    ae instead of the ligature but certainly not the other way round. But to    be honest, I personally never used such a word in German.       Therefore the reason to make ae and its ligature variants would be    purely semantic. And then we have a problem: how to decide which variant    relation is stronger? If all are visual, it's (more or less) easy to    decide which visual similarity is stronger, but there's no metric to    compare visual and semantic similarities with each other.       Cheers,       Michael       --    ____________________________________________________________________         |       |         | knipp |            Knipp  Medien und Kommunikation GmbH          -------                    Technologiepark                                     Martin-Schmeisser-Weg 9                                     44227 Dortmund                                     Germany            Dipl.-Informatiker          Fon:    +49 231 9703-0                                     Fax:    +49 231 9703-200         Dr. Michael Bauland         SIP:    Michael.Bauland at knipp.de         Software Development        E-mail: Michael.Bauland at knipp.de                                        Register Court:                                     Amtsgericht Dortmund, HRB 13728                                        Chief Executive Officers:                                     Dietmar Knipp, Elmar Knipp    _______________________________________________    Latingp mailing list    Latingp at icann.org    https://mm.icann.org/mailman/listinfo/latingp   _______________________________________________
Latingp mailing list
Latingp at icann.org
https://mm.icann.org/mailman/listinfo/latingp _______________________________________________
Latingp mailing list
Latingp at icann.org
https://mm.icann.org/mailman/listinfo/latingp

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mm.icann.org/pipermail/latingp/attachments/20190212/4b9ca0cb/attachment-0001.html>