Meeting minutes
<AvneeshSingh> https://
wolfgang: today we are speaking about language tagging. The accessibility task force have prepared a document and we hope to have some feedback.
GeorgeK: in epubtest.org under non visual reading we have a test for switching languages, it's simple, just test a span in a paragraphe, once in french, german, spanish and english. The goal is to see if the screen reader changes the voice. NVDA allows to set an option for switching or not, there is also an opion for switching to regional variations like french from canda, etc.We don't check from dialects, only main languages.
AvneeshSingh: that triggered us to work on best practices, and realised this is a complex space, we need discusions before writing best practices. This is the goal of this discussion paper. What behavior do we expect from assistive technologies (AT), what powers should the user have (settings), what is doable with authoring tools we have today? How many words should be triggering a specific language tagging? How do we deal with minor or
old languages that have no tts voices?
AvneeshSingh: aside from the document the persons involved have raised some issues, we raised questions, we want your insights.
gpellegrino: working with publishers we see there are many regulations refering to wcag. We've been tackling the images alt texts for years; Now we see that language tagging is the raising burden.
<AvneeshSingh> questions: https://
gpellegrino: how much is it important for end user? Because yes it is a burden.
PaoloCasrini: i believe language tagging is important feature, it improves the semantic quality and enable AT. But also, I don't think that it is a real blocker for AT users. Incorect language tagging reduce the experience but in many cases in does not prevent usage of the book. As publishers, it is a big burden to produce at this moment, so we need to be able to wisely determine which are blocking and need to be done from which would enhance the experience but are not blocking.
… without guidance we risk to claim disproportionate burden.
Bill_Kasdorf6: as a consultant I see diverse books, medical scientific, etc. Good news, I see more and more ebooks that have language tagging. Many ebooks already have italised sentences in other languages, that makes them easy to identify. Also, AI is changing the game, if it is not used by publishers it is used by producers.
GeorgeK: because of the growing popularity of read aloud, it's not just about AT, it's also a topic for readinng systems now (RS)
gpellegrino: on AI, it helps, but need a lot of attention, we've seen hallucinations between latin and italian, and on smaller languages.
mgarrish: i agree that a book with not all languages tagged may not be inaccessible, that's part of the wcag, the rule let place for human interpretation. It's difficult to state what is accessible and what is not. Not everything is gonna be perfect.
sueneu: it makes me think about the concept of gracefull degradation. We don't know how to pronounce ancient greek. We should not be too strict and let space for this gracefulll degradation.
DaleRogers: that challenges how we see conformance.
gautierchomel: maybe we can take one or two questions of the document and adress them as topics.
<sueneu> +1 @gautierchomel suggestion of round tables for each issue
PaoloPancaldi: I am a developper, I spent many hours for tagging and it is complex, expensive in time. Whatever technology used, it require revision. It's usefull, but we need a more detailled guidance on to which point we should perform.
GeorgeK: my personal opinion is that unless it is a langugae training book, I don't think the absence of language tagging will make it inaccessible. When I turn on language switching in NVDa and read emails from Avneesh and Richard the voice changes to english from India and English from Britain, I prefer not to have this difference. But with a language learning book, yes I want to have a voice shift.
<AvneeshSingh> daisy/
AvneeshSingh: which question should we take?
Bill_Kasdorf6: my publisher clients are trying to tagg as many things as possible, evene if it says "voila" in an english book, most will want it prounounced correctly. Publishers are involved in quality.
wolfgang: I see that too, in German, we want the correct pronounciation, like the correct spelling.
gautierchomel: trying with 9.6.1. What should be considered a "part" for language-switching purposes?
to me it's an editorial decision, it depend on the book, collection, publisher.
gpellegrino: for us it is 5 words, if it is the same alphabet but only one word if it is not the same alphabet, in that last case, that will block. This are our rules for certifying
GeorgeK: i think it also depend on language groups, maybe in cases it would trgigger different feedbacks.
mgarrish: we cannot modify wcag.
Bill_Kasdorf6: there's gonna be judgement involved, but many time those tagging have to be done.
gautierchomel: the langugae group things is real, some africans french speaking countries are happy with french from france voices, while others cannot use it because they feel it's too much ununderstandable for them.
gpellegrino: we cannot change wcag but having some precision is what we ask. We did that for EPUB accessibility. I agree to have the bar high but the reality is far.
gautierchomel: wcag does not define parts, we may ask them why.
gautierchomel: to me that's because it's an editorial responsability.
mgarrish: we certainly have to ask them before publishing anything.
<AvneeshSingh> Links to all questions: daisy/
<AvneeshSingh> daisy/
<AvneeshSingh> daisy/
<AvneeshSingh> daisy/
<AvneeshSingh> daisy/
vincentml: I feel we should tag anything and let reading systems, TTS engine empower users.
AvneeshSingh: I added all links to discussions, so they will be in the minutes, I encourage everyone to pass the links to colleagues so we can have the wider possible feedback.
<Bill_Kasdorf6> +1 to Avneesh
<jsaiya> +1 to gautierchomel
AvneeshSingh: also I agree we could have other plenaries on the topic but focusing on specific issues or targetting specific audiences.