If that is the case, then why not map them to role="none" unless they are made focusable and then you map them to the native host language semantics. A role of none will not hide the content within the container having doc-pullquote unless they are structural elements associated with the
Or, do you really want to hide the element and all the content of a pull quote? Note: if you do this then you have to deal with an author putting tabindex on it and child content. 

Rich Schwerdtfeger
----- Original message -----
From: "Siegman, Tzviya - Hoboken" <tsiegman@wiley.com>
To: "jcraig@apple.com" <jcraig@apple.com>, Joanmarie Diggs <jdiggs@igalia.com>
Cc: Rich Schwerdtfeger <richschwer@gmail.com>, Alexander Surkov <asurkov@mozilla.com>, Bogdan Brinza <bbrinza@microsoft.com>, "DPUB-ARIA (public-dpub-aria@w3.org)" <public-dpub-aria@w3.org>
Subject: RE: doc-pullquote
Date: Thu, Oct 27, 2016 12:18 PM
The reason we have suggested that pullquotes be hidden in some way is that they duplicate content.

James, based on discussions with users of AT, we concluded that most will find pullquotes irritating and/or confusing because they duplicate existing content. So, they might provide emphasis, but we'd like the default for AT to be not displayed. How can we make that work?  

Tzviya Siegman
Information Standards Lead

-----Original Message-----
From: jcraig@apple.com [mailto:jcraig@apple.com]
Sent: Thursday, October 27, 2016 12:57 PM
To: Joanmarie Diggs
Cc: Rich Schwerdtfeger; Alexander Surkov; Bogdan Brinza; Siegman, Tzviya - Hoboken; DPUB-ARIA (public-dpub-aria@w3.org)
Subject: Re: doc-pullquote

Sometimes they aren't purely decorative. At a minimum, they provide emphasis. Arguably, they could be navigable, too. Similar to headings or images, but probably not recommended to subclass landmarks.

> On Oct 27, 2016, at 9:17 AM, Joanmarie Diggs <jdiggs@igalia.com> wrote:
> But my understanding is that pullquotes are purely decorative and ATs
> shouldn't double present them. Native host language semantics won't
> necessarily buy us that, right?
> --joanie
> On 10/27/2016 12:05 PM, Rich Schwerdtfeger wrote:
>> Not mapped only means that the element is not mapped. Hidden should
>> apply to the element and its subtree unless a descendant is
>> overridden by the element. So, I think a better solution would be to
>> not to expose the role at all and allow it to default to the native
>> host language semantics. In this case it does not matter whether it has focus or not.
>> In fact, I think it would be best to state that in the mapping spec.
>> that defines what it means to not be mapped. We should say that it
>> defaults to the native host language semantics and the ARIA role is
>> not exposed.
>> Would this not be a better approach?
>> Rich Schwerdtfeger
>>> On Oct 27, 2016, at 9:24 AM, Alexander Surkov <asurkov@mozilla.com
>>> <mailto:asurkov@mozilla.com>> wrote:
>>> Thinking more, aria-hidden='true' might be not that bad, even with
>>> its ambiguity of the usage and implementations in the browsers. If
>>> you say 'not mapped', then I think you should say if this is applied
>>> to elements from its subtree, also you may need to clarify the edge
>>> cases, for example, what happens, if the focus goes into 'pullquote'
>>> element.
>>> Having said I would probably prefer if that was the author's
>>> responsibility to hide the element from AT, either by moving it
>>> offscreen or putting aria-hidden='true' on it.
>>> Also I don't quire understand why would we need to have 'pullquote'
>>> role in the first place. If pullquote doesn't have semantics, then
>>> it doesn't seem worthwhile to have ARIA role for it.
>>> Thanks.
>>> Alexander.
>>> On Thu, Oct 27, 2016 at 9:34 AM, Rich Schwerdtfeger
>>> <richschwer@gmail.com <mailto:richschwer@gmail.com>> wrote:
>>>> Based on this discussion:
>>>> https://lists.w3.org/Archives/Public/public-dpub-aria/2016Oct/0029.
>>>> html
>>>> I want to change the dpub-aam mapping to not mapped for all
>>>> platforms. That includes (not xml-roles values for linux and
>>>> Windows IA2 mappings).
>>>> It means
>>>> that the Mac mapping would go form group to not mapped.
>>>> Any disagreement?
>>>> Regards,
>>>> Rich
>>>> Rich Schwerdtfeger