Comment on a11n, l10n and i18n

[Omikhleia]

a11n, l10n, i18n accessibility (e.g. images must provide alternative text representations for blind people), localization (e.g. numbers should be representable as ‘10,000’ as well as ‘10.000’ depending on the locale settings), internationalization (e.g. support for page size ‘letter’ as well as ‘JIS A’)

Interestingly, it doesn't mention respecting proper language-specific typography... Doesn't it matter too? Numbers depending on "locale" are sort of easy with ICU. That's just a tiny part of the problems at stakes, as the typographers here certainly know. Non exhaustive examples:

• Special spaces in French around "high" punctuation or quotations marks... And no, these are not fixed-sized en- or em-spaces...
• English spacing rules after some punctuation (e.g. what TeX calls nonfrenchspacing, active by default, but it could have been called englishspacing and not made a default, had Knuth thought a bit more about it, heh! Modern solutions, and usage, tend not to apply these rules any longer, with (La)Tex being an exception, but anyway...)
• Proper bulleted lists = yep, not all countries/languages favor the same "bullets" in the same order...
• Collation and proper sorting in indexes, automated glossaries and bibliographies = yep, i18n goes beyond selecting a paper size or how to format numbers (and where, by the way? Ordinals, pseudo-ordinals, decimals, durations, amounts... Even ICU renounced here, it seems, to catch it all ;)
• ...

I am not sure from which country the author(s) of this document come, but there seems to be a slight bias, probably unintended if I get it right, toward the simplest aspects of digital typography made right ;-)

Also, I'd suggest splitting accessibility apart, if the author(s) of this document believe it's a key item. It's sort of unrelated to l10/i18, and there would be much more to say about it, really. Images are a visible aspect -- no pun intended -- of the problem, but there are plenty of other issues for vision-impaired people. Ask a few people in this situation and start with considering multi-column documents, page headers and footers, pull-quotes, insets, maginal notes, etc. = how current solutions work for them... or not...

[tajpulo]

Thanks a lot for the input!

I am neither new to the field and the choice of “simplest aspects of digital typography” is exactly my goal. My intention is really to get a list of software doing at least the bare minimum to be included. And the next steps of the website is to guide the user through a ranking of comparable features. And then it should get obvious that Teχ (pretty much at the top of the website) is more valuable than [e.g.] Paged.JS (at the bottom of the website) in your typesetting stack. But I am annoyed that people are not even aware of solutions like typst, speedata publisher, and SILE. So I want to provide some overview with this website.

First of all, I need to get the initial list right and thus the criteria. The list - you mentioned - makes sense, but only captures a tiny fraction with a bias towards Latin script. I think “hanging punctuation” might be an interesting requirement if we want to use some basic typographic requirement at all.

And you might be surprised - if you don't know - how few software packages ending up on this website use ICU :wink:

Anyhow, your remark on a11n made me think. I don't have strong opinions on merging/separating these criteria. [My first draft included all these aspects in six points :innocent: ] But I get your point that l10n is much closer to i18n than any of them is to a11n. Hm.

[Omikhleia]

I don't have strong opinions on merging/separating these criteria. (...) But I get your point that l10n is much closer to i18n than any of them is to a11n

To better explain my point if need be:

First of all, accessibility covers a lot of different things and needs (addressing vision-impaired people, but also people with hearing, mobility or cognitive impairments). Since the focus here is on typesetting system and mentions alternate text on images, I only focused on vision impairments. This might be somewhat implicit, but if it stays it you requirements, it might have to be clarified however (which part of a11n is considered a requirement indeed?)

This said, accessibility for vision-impaired readers stands a bit apart from localization/internationalization, and meets a lot of extra challenges. As I noted above, alternate text on images is only a very tiny aspect of it. Would some of the listed solutions enforce having alternate texts on images (thus ticking a checkbox here?), it still wouldn't mean much...

I can't speak in general, I am not a specialist and don't want to be too personal here. But I just happen to know some people in that situation. When your text-to-speech assisting software starts reading aloud URLs (eeek!), or page headers and page numbers at each page break, even in the middle of a sentence (doh!), or worse, to read multi-column text or tables without any sensible logic (ouch!), it's getting pretty bad for them... Not to mention the simplest footnotes, if they are purely rendered visually (i.e. a footnote call number) with no way to navigate from the text to the footnote content and reciprocally... However nicely a text is typeset, say, to a raw PDF, it may look real great but be impractical for people with disabilities.

There are some ways to improve this: properly done "Tagged PDFs" may help a bit, for instance... Though it's not widely adopted and properly implemented in readers, and has a number of issues too.

So to keep it simple here: a11n really stands apart from language-specific aspects (i18n or l10n), it seems to me. It could still be a criteria for typesetting systems (in the sense that the output then has to help with interpreting the logic of the text flow), but I'm afraid all of the listed software are going to utterly fail the test.