The idea is that, just as you trust sighted users to parse the italics, you need to trust the screenreader users as well.
By semantic information, we mean use the tags with meaning instead of look whenever possible. <q>, <em>, <blockquote>, and <cite> might all be rendered differently by a screenreader; use them to give the screenreader a chance to cope with them. If you want a specific look and feel but, say, BLOCKQUOTE is appropriate and yet gives you a look and feel you don't like, style the tag with CSS to appear to sighted users as you like while still using the semantically right tag. But if that doesn't work, use <i> if nothing else applies. Just as sighted readers figure out how to parse italics (I know the ones used in your comment here are quoted strings, not emphasis or dream sequences), screen reader users can figure out the same thing.
no subject
By semantic information, we mean use the tags with meaning instead of look whenever possible. <q>, <em>, <blockquote>, and <cite> might all be rendered differently by a screenreader; use them to give the screenreader a chance to cope with them. If you want a specific look and feel but, say, BLOCKQUOTE is appropriate and yet gives you a look and feel you don't like, style the tag with CSS to appear to sighted users as you like while still using the semantically right tag. But if that doesn't work, use <i> if nothing else applies. Just as sighted readers figure out how to parse italics (I know the ones used in your comment here are quoted strings, not emphasis or dream sequences), screen reader users can figure out the same thing.
More on those different tags, explaining the differences.