Re: [css-text] I18N-ISSUE-333: 'letter-spacing' and Arabic

On 5/30/2014 11:08 AM, John Hudson wrote:
> On 30/05/14 5:08 AM, Richard Ishida wrote:
>
>> Is it really expected that implementations decompose optional ligatures
>> when 'stretching' Arabic text? Are we just making assumptions here, or
>> is this based on some typographic tradition?
>
> The whole notion of 'required' and 'optional' ligatures in Arabic is 
> extraneous to the script tradition. Indeed, a ligature is a particular 
> (obsolete) technological mechanism (a glyph/sort representing more 
> than one character), and a function to disable 'optional ligatures' 
> presumes that digital fonts provide ligatures that could be disabled. 
> An increasing number of digital fonts do not (and not only for Arabic) 
> and instead use contextual variants to shape text, even to shape 
> 'required ligatures'. Obviously, you can't 'turn off ligatures' if a 
> font contains nothing that indicates a particular typeform is a 
> 'ligature'.

Interestingly there are (or were) Latin orthographies where these 
concepts rose to the level of orthography. German orthographies of the 
20th century would have rules that prohibited the use of ligatures based 
on the meaning of the word (whether letters where adjacent because of 
compounding made a difference) so no rendering system could consistently 
apply ligatures for the same letter context.

In particular, the Fraktur style of German knew several obligatory 
ligatures. These were not left to the discretion of the typesetter. No 
rendering system attempting to reproduce that style could simply "turn 
off" all ligatures.

Optional ligatures were used or not used depending on the requirements 
for justifying text. Fraktur, typeset in the traditional way is not very 
forgiving of large inter-word spacing, except around punctuation.

In 'letter-spacing', which was absolutely the standard way to provide 
emphasis in that style, obligatory ligatures were kept together (no 
space added between their constituent part, only around them), but 
optional ligatures were not applied (all the individual letters had 
space around them).

I find it unsurprising that typographers aware of such systems would 
apply these concepts to Arabic, whether that is the best possible 
description or not.
>
> Given that what is referred to as 'ligation' in Perso-Arabic scripts 
> is actually just the usual contextual behaviour of specific sequences 
> of letters according to the conventions of particular styles of text, 
> trying to disable them at a mechanical level, regardless of the style 
> of text or how it has been implemented in a given font.
>
>> (I know that high end justification systems may create ligatures as part
>> of the justification process, but) from what I've seen it doesn't
>> necessarily follow to me that there is always a logical first step to
>> the stretching of arabic text that dismantles optional ligatures.
>
> Right. If it is a viable option, it would be so in the context of a 
> particular style and a particular font implementation, not something 
> to be done at a higher level independent of that context. It's the 
> sort of thing that the OpenType JSTF table could conceivably assist, 
> but work needs to be done to define both font and software 
> expectations in this regard.
>
> JH
>
>
>

Received on Friday, 30 May 2014 19:03:11 UTC