I can't comment on using Zotero with voice recognition software. On opening PDF files in Firefox or outside of Firefox, see the "launchNonNativeFiles" hidden preference on the hidden prefs page. This allows you to force PDF attachments from Zotero to open outside of Firefox.

For sites like ProQuest that force you to read PDFs in the browser, you can actually make Firefox open them in Acrobat, or make Firefox download them. See this page. But even better, you can save the reference in Zotero, which will in most cases download the PDF for you, and as soon as the PDF is attached to an entry in your Zotero library you can open it outside of the browser (if you set the above hidden pref).

The CHNM receives substantial funding from Federal agencies such as the NEH and IMLS. These agencies require grantees (such as Zotero) to conform with federal regulations on accessibility. The IMLS, for example, requires grantees to comply with the Rehabilitation Act and the ADA. If Zotero is not accessible to people with disabilities, then the CHNM may in fact have violated the terms of its grants. At the very least, Zotero provide information on accessibility to users.

Zotero has always been designed with the needs of users with disabilities in mind.

For example, in order to assist users with disabilities, all mouse-driven Zotero functionality should also be accessible via keyboard input. To the best of our knowledge, any voice-driven control should also be able to access this functionality.

If there is specific functionality that users are unable to access due to a disability, we are committed to being responsive to those users' needs. Since Zotero is a community-driven project, as others have already noted, it is possible that some features have been incrementally added that are not, to use the example above, keyboard accessible. Although the Zotero project has continued to grow long behind the limits of its initial grants -- the IMLS funding ran out over two years ago -- we are of course committed to meeting the needs of all of our users.

Now, on the specific problem that you've encountered, what exactly is happening when you use Zotero? Are you able to access the individual text input fields in a given item? Is it that you cannot enter text into those fields? If not, are you able to enter text into other input fields in Firefox, say, for example in a web form? Let us know what's happening, and we'll go from there.

after I locate the mouse in the field and dictate text, the computer gets hung up. Sometimes this is for 5 to 10 seconds, usually it is indefinitely.

Are you referring to notes or fields? (Or both? But they're implemented completely differently, so that seems unlikely.)

If you're referring to notes, are you able to use the demos on the TinyMCE site? (Zotero embeds the TinyMCE editor for rich-text editing.)

Either way, this doesn't seem to have anything to do with PDF handling.

One problem here is that database vendors, in the interest of protecting their copyright, strip their indexing when documents are downloaded into Adobe --it is therefore impossible for disabled users to keyword search documents like able-bodied users can when the document opens within a browser.

I'm not sure what you mean by this. How would a disability affect whether you can search by keywords?

I'm pretty sure the problem is with Zotero, not my computer set up.

We're happy to help try to debug this, but be warned that Zotero doesn't really have much to do with stuff like text input. Any system that's properly generating text should in theory be indistinguishable from keyboard input, and the Mozilla application framework should do the rest. (For example, a properly functioning barcode scanner converts a bar code into a string of numbers and simulates keyboard input for those numbers. You can use one with Zotero's Add Item by Identifier functionality, but Zotero certainly doesn't know anything about barcode scanners. All it has is a text box.)

But if there's anything we can do on our end to improve accessibility, we'll of course try to do it.

Dan: Notes work fine. Fields do not. I checked on TinyMCE.

I should've been more specific: having a disability doesn't necessarily affect whether one can search by keywords, but using assistive technology at the computer does. Here's how it works. Serving up PDFs in browser windows conflicts with most of the assistive technologies people use. Therefore, the accessibility guidelines in Adobe require users who use ANY assistive technology to open all PDFs within Adobe rather than a browser. Now, when able-bodied users perform a keyword search in a fulltext online database such as EBSCOhost, they can click on individual items and view the PDF within a web browser. Usually, one can search within that PDF. The most obvious utility of such a search is instantly locating where your initial search term occurred in the document. However, what you are viewing in the PDF browser is an image--when I download that image to my computer and open it within Adobe, as I must do in order to use voice recognition software, it is no longer searchable. The vast majority of database vendors do not include an embedded text layer in the image (JStor does): because they strip the text layer from the PDF, they discriminate against users who rely on assistive technology to access the information in their databases because such users can't access the PDF within a browser window. Here is what Adobe has to say about this issue:
“The first step in making scanned documents accessible is to perform OCR on the scanned page image. Acrobat has an OCR text recognition feature that allows you to apply OCR to the scanned pages.
The resulting PDF files contain computer-generated text, which is necessary for making the file's information accessible via screen readers and other assistive technologies. You may need to further process the files at this point by using the accessibility authoring tools in Acrobat to add structure (tags), alternate text for graphics that appear in the file, and accessible form fields if applicable. You may also need to adjust the reading and tab order for interactive PDF file components.”
More here: http://www.adobe.com/accessibility/
What database vendors do is strip the OCR from their documents when they are downloaded into Adobe, making them inaccessible to the very users who rely most on these features.

The fact that opening a PDF file outside of Firefox is within Zotero’s “hidden preferences” ("launchNonNativeFiles") is an example of the kind of cultural shift that needs to happen. While many might feel that the choice of opening a PDF within a browser or not is merely a preference, users who rely on assistive technology can be completely locked out by hiding this kind of information.

Sean: if the Zotero “has always been designed with the needs of users with disabilities in mind,” then it should be upfront about that fact and provide pages dedicated to accessibility that explain manipulation of font sizes, keyboard input, voice recognition, and use with screen readers.

Because Zotero is hosted at a major US university, testing the software with assistive technology is not a financial barrier. You could, for example, contact the Office of Disability Services at George Mason and mess around with Zotero using the most common technologies used by disabled academics. I suspect you even could get disabled faculty and students to do such testing for you.

Respectfully, I am not wrong about PDFs. I may be wrong about the technical details (I am a historian), but I am not wrong about the usability issues. Use Proquest for 19th-century documents and see if there are no issues.

Now, when able-bodied users perform a keyword search in a fulltext online database such as EBSCOhost, they can click on individual items and view the PDF within a web browser.

This sound like individual page images displayed in the browser, not PDFs (although they look the same). It seems entirely possible that EBSCOhost could have a text layer that it uses to decide which pages have matching text, and that that text layer is _not_ included in the PDF files that EBSCOhost serves. Adamsmith is likely correct that there is no difference between opening PDF in a browser plug-in versus in an external program. There may, however, be metadata that article databases don't ever provide in a downloadable form. This is not something that Zotero can do much to improve, but maybe EBSCOhost can be pressured to release the OCR'ed text for accessibility reasons.

Dan. I still think there are differences in usability for people using assistive technology when searching, but let's shelve this discussion until I can give you specific examples (in any case this doesn't affect getting Zotero compatible with Dragon). Gathering specific examples is difficult for me --all the computers I work on are now set up to bypass the default settings in Firefox and Adobe, and, they are usually in enclosed rooms in the libraries I work in. I will need to get somebody to run identical searches on a “normal” computer, and perhaps talk with people who use screen readers. What people seem to assume is that I can easily compare my experience with that of a person not relying on assistive technology. Just as it is a challenge for Zotero’s developers to “go the other way” and test Zotero using assistive technology, I can't easily replicate using these tools without assistive technology (I have limited use of my hands/developers may have limited access to computers equipped with accessibility software and hardware, and there is some investment of time in learning the software).

I use OCR extensively. My point is that disabled users should not have to go through a resource and time consuming extra step in order to access information they retrieve from database vendors when able-bodied users enjoy instant access. In my case, this involves saving files to disk and OCRing files on another computer as the computers I use are not powerful enough to run Adobe’s OCR function in the background while I am working. For me, this is not a big a deal (I am not visually impaired, nor do I need text to speech to understand what I'm reading). But for people who use assistive technology that relies on OCR, this policy is a huge problem. (What database providers don't OCR their images? If the images are not OCRrd, they are not searchable, right?) I have contacted some of the big database vendors about this issue, but they simply don't reply. Librarians are generally not informed about digital accessibility issues, and my experience when I have not been able to get databases to work with Dragon in my library has been that they refer me to either IT or the database vendors themselves. The database vendors don't seem to test their products with assistive technology in the sort of real-life computing environments that academics work in. The lack of information about how to get this stuff to work has serious consequences: when I started using Dragon NaturallySpeaking, it took a few months before I was able to conduct library searches using voice recognition (I was teaching and in the middle of coursework). Moreover, I had to schedule demonstrations with librarians and IT staff to prove to them that the computer setup they had provided didn't actually work using databases like ProQuest and EBSCO. Of course, getting database vendors to not withhold their indexing would benefit all users.
Here are ways Zotero can help.

1. Make Zotero accessible: start with supporting Window-Eyes, JAWS and Dragon Naturally Speaking. Mozilla already does this (http://www.mozilla.org/access/) and identifies accessibility as a major priority. Zotero could probably get funding for this. But part of getting funding would be identifying the problems that researchers who use assistive technology face.

2. Help identify those problems by providing a space for working academics to evaluate the accessibility of the online resources Zotero links with. This would enable academics to develop a set of “best practices” for serving up digital information. My experience is that JStor does this better than anybody else, and one way to start would be to figure out exactly what they do and rank other databases against them. We might assume that such practices had already been developed, especially given the not insignificant amount of funding devoted to assistive technology at U.S. universities. But the barriers for disabled academics are extremely high, and my guess is that there are few people actually utilizing these online resources. For example, at my large public university only about one out of 1000 doctoral students have self identified as disabled to the University, and despite the high incidence of RSI, all the academics I know who admit to using Dragon NaturallySpeaking I have met online.

Regarding your specific questions:

I don't think it is Firefox that is freezing . When I input text into the title field, the text almost never drops into the field. Dragon is working, as is Firefox (I can work in another window, although the moment I click anywhere outside of the box in which I am trying to enter text, the operation in Zotero stops). I am pretty sure this is not a resource issue: there is a small bump in the CPU and memory usage when I dictate into Zotero (this would be much larger on a more mainstream computer, but the Nehalem chip handles Dragon incredibly well), but nothing that could hang the computer up.

I have gotten the info section to work a number of times, and when it does, the text drops right in the box like it should and I can navigate around the different fields, changing text at will. Once it stops working, though, it will hang up indefinitely. Getting out of the Zotero seems to help “reset” it, as does dumping in a new record.

Tags work great: the text drops almost instantaneously into the box.

Text does not drop into the quick search box.

But, when I go to the “related” tab of a record and click “add,” the search box in the pop-up window DOES work. Also, the searching by ISBN/PMID function works fine.

I get that. Searching is one thing, but "reading" is another.Without the hidden text layer a disabled user can't "read" the document until it has been OCRrd. You and I can read it on-screen. What you don't understand is that many disabled people can't access the information in PDFs without the text layer: assistive technologies take that text layer and convert it to a format that is accessible. Not all people can see. Not all people visually process text. Therefore, images (a collection of JPEGS or TIFFS) are not accessible. Commonly accepted accessibility standards dictate that images on websites contain an embedded text description for visually impaired users. (Zotaro does not comply with the standard). When a database vendor has clearly OCRrd a document (because it pops up from a keyword search related to information in the document itself rather than the abstract) and then they strip that layer they are making that document inaccessible to users that do not visually process text or who need technical assistance to do so.

Dan - I think there are two separate issues that have confused the thread.

One of them is related to Zotero - the way it performs with standard accessibility software.

The other, the way database providers treat hidden text, has nothing to do with Zotero - I think everyone understands that - but if I understand sean correctly he used it as an example for the types of issues researchers with disabilities face.

I think number two is pretty pointless to discuss here, because no one present will be able to do much about it, while I think it would be great if issues related to number 1 could be removed.

So, regarding text fields, frankly this sounds like a problem with either Dragon or Firefox. The right-pane text fields are created dynamically, which perhaps could cause issues, but tags are implemented the same way, and you say those work. The quick search box is just a text box, identical to the main Firefox search box, so I'm not sure why that wouldn't work.

Are you able to use Dragon in all other non-web text boxes in Firefox? The Firefox location and search bars? The search box in the "Get Add-ons" pane of the Add-ons window? The search box in the Organize Bookmarks window? How about text boxes in other Firefox extensions?

Have you tried disabling all other Firefox extensions? Any other patterns you notice?

Can you provide a Report ID after reproducing the problem?

Until Zotaro does accessibility testing and incorporates universal access into its design process, I am not particularly interested in what you think this problem “sounds like.”

I've not really followed this thread, and am fairly ignorant about accessibility issues. But I suggest you rethink how you approach engaging this forum and the Zotero developers who contribute to it. Comments like this are simply not cool or constructive.

My apologies for being abrasive. Please understand that virtually every time I have ever interacted with a librarian, IT person, software maker, or database vendor about trying to fix a problem accessing electronic databases while using Dragon, they have basically given me some version of "the problem is surely with [insert other software maker, or the database vendor, or your computer configuration]."

Dan: [am I] able to use Dragon in all other non-web text boxes in Firefox?
yes.
The Firefox location and search bars?
yes
The search box in the "Get Add-ons" pane of the Add-ons window?
yes
The search box in the Organize Bookmarks window?
yes
How about text boxes in other Firefox extensions?
I don't use any other extensions that require input.

Sean: I disagree. I think that there would be tremendous value for developers to test Zotero using various assistive technologies. Of course I want to use Zotero--even with my limited experience of messing around with it, I can see all kinds of potential not possible with Refworks and Endnote. But at the end of the day, I would like to see Zotero work for lots of disabled people using lots of different assistive technologies: I don't think you will be able to make Zotero accessible without doing some testing on your end. But I don't know. The people who will know are the folks at Mozilla, as they have gone through this process and my experience with using Dragon In Firefox has been decent (there are things that could be better, but I can use it).

The report errors box remains gray.

I disabled the other extensions, and something has seriously messed up my sound system. First, Dragon NaturallySpeaking crashed. After multiple restarts, I have got it working, but the quality of the sound is degraded (Dragon is telling me that some other application is now hogging the sound card). I will fuss around with more by turning on extensions one by one and seeing if I can get the sound quality back up to speed. (I am a luddite, and I also disabled plug-ins when I did the extensions because that was the example that showed up on Mozilla's website, but after re-enabling the plug-ins I'm still having problems with sound quality.)

As of this morning, I'm not always able to insert text into the notes field. but if I open a separate browser window, the text drops in right away.

Keyboard input always works, whether or not Zotero is receiving input from Dragon.

Returning to more general issues. . . I think what rubs me the wrong way about these discussions is the lack of recognition that accessibility is a civil rights issue. When a software tool, especially one that is being adopted extremely quickly in the academic community, doesn't work with commonly used assistive technology it should not be treated as just another bug. Moreover, if Zotero is going to make the decision to not support such technologies, it should be upfront about it and clearly state on its website that researchers or institutions who need such support should use one of the commercial products.

I think what rubs me the wrong way about these discussions is the lack of recognition that accessibility is a civil rights issue. When a software tool, especially one that is being adopted extremely quickly in the academic community, doesn't work with commonly used assistive technology it should not be treated as just another bug.

@uptownsean: Your audience is listening to you, you might try listening back. Those who designed and built Zotero are committed to the community, including your good self. They take pride in their work. The suggestion that they take flaws in the fabric of their creation lightly is offensive. But then you probably know that.

Have you considered the possibility of getting someone from Dragon involved in the conversation? Here's the thinking. Bugs (even very important ones) are tracked down by checking the behavior of everything in the chain. As Dan Stillman has pointed out, Firefox and Zotero present standard targets for text input, and nothing appears to be broken there. Dragon is a closed-source proprietary product, and it is not possible for outside developers to see precisely what it is doing when it interacts with Zotero, Firefox or the operating system. No one knows whether Dragon itself is the source of difficulty here, but one of their engineers might well be able to identify the issue very quickly, to everyone's benefit.

Something to consider, anyway.