Here's an NVDA in-line citations filter

If you read a lot of scholarly literature (e.g., liberal arts or science college students), you're doubtless familiar with the pain of having to listen to long strings of in-line citations like "(Gwen, Einstein, 2008 [1905]; Calvin and Hobbs 1989: 57)" etc., etc. After 30 years of putting up with this, I finally realized that, for how ever many years now, NVDA has a speech dictionary that accepts regular expressions. These are extremely powerful and inscrutable strings that match very flexible and complex patterns. So, I wrote a regular expression to recognize lists of two or more in-line citations, then plopped it into a speech dictionary entry that replaces them with "CITATIONS". Accessible e-readers like Voice Dream and Listen2 have a similar feature. If you need it, you know it.

Here's how to do it:

1. Go to NVDA > Preferences > Speech Dictionaries and pick whatever scope you want it to be active ("default" is global, "Voice" is for the current voice, and "Temporary" disappears on restart).
2. Add an entry and paste the following into the "Pattern" field:

\([^)]*?(19|20)[0-9]{2}[;:\],a-z]{1,2}[^)]+?(19|20)[0-9]{2}.*?\)

3. Enter whatever you want into the replacement field. Text in the "comment" field will appear in the dictionary entries list. "Case sensitive" doesn't matter.
4. Change "type" to regular expression.
5. Hit ok, then ok again.
6. Whip up a test document or open something heinous to read.

What it actually does is match text inside parentheses that contains at least two dates from the 20th or 21st century, with optional text of any kind before or after the dates. The first date also has to be followed by the kinds of things that mark it as a citation in APA, Chicago, or MLA style, including a colon before a page number, semicolon, lowercase letter (2025a, 2025b), or right bracket (for a secondary publication date).

It's certainly not bulletproof, but it's working plenty well enough for me. Hopefully, it works for you. I do a lot of scholarly reading, so find it life-altering.

A couple of things to point out: If a word in the parentheses has been marked as misspelled, NVDA won't match it. Also, there obviously won't be a match if the citations break across lines, so this works best when reading by sentence or paragraph. And, of course, moving into the citation list by word behaves normally.

For a truly awe-inspiring application of regular expressions, check out the Markdown Navigator addon by Cary Rowen.