A screen reader can emphasise (em) and stress the importance (strong) of words, just to mention a couple of elements that are used for better semantics. A designer can target these semantic elements and stylistically present them to offer better visual semantics.
Surely, a screen reader should do the right thing when it meets phrases such as, The SAS soldier visited the NAAFI to meet his fellow troopers from 'The Regiment'.
Here we have an abbreviation, an acronym and a definition. Each should be marked up differently and titled, 'Special Air Service', 'Navy, Army and Air Force Institutes' and 'The name used, affectionately, by members of the Armed Forces to refer to members of the SAS regiment' respectfully.
A screen reader would spell out the letters 'S' 'A' 'S', say the word Naafi and possibly make use of the titles, especially for the definition; 'The Regiment'.
A designer could also stylistically present each of these semantics to visually identify them, if desired.
To complete our progressive enhancements, a JavaScript programmer would be able to parse the markup and change each of these into a link to Wikipedia for a free glossary of terms.
Obviously, you would need to make sure you got these the right way around, because you would not want to call an SAS man a 'sassy'!

This is just as, in fact more, important than some of the elements that are part of HTML5 currently; (i) and (b) for example.
It is hurting me and many others I have discussed this topic with, that we have lost the acronym element.
I have heard all the arguments for and against but none have convinced me that ACRONYM should not be one of the important elements of the HTML5 semantics.
Please do the right thing and rectify this error of judgement.
Thank you.