Avoid reading superscript, e.g. citation (reference) numbers on Wikipedia etc.

Is there a way we can exclude citations starting with numbers in reduced font size?

Following is the sample: 
 
Paul Ree exemplifies a tendency to place present uses of moral concepts — ‘punishment’ for what deters the egoistic, and ‘good’
²⁶ GM II. 12, on which, see also Leiter (2002: 168, 198).
²⁷ On Ree’s account of punishment, see Ch. 5 below.
used as praising the unegoistic—at the origin of morality.
 
This is how the text is being read. After reading 'good' the actual sentence has to be continued from 'used'. I think it is bit difficult to make it leap to 'used' accurately. But if we can just skip the respective sentence which starts with a number in lower font, it will be good. I can send the full PDF if it helps

Sometimes when I encounter this in ebooks, the citations are at the end of the chapter, and I can skip them by manually scrolling down to the bottom and choosing the next chapter option. On some articles, this can be quite a long scroll, however. And nonchaptered ebooks do not have this functionality. Also, OCR of certain books I've read nondigitally would have line-splits as the text wraps around same-page citations... newer made-for-screens versions have no reason to do that, but everyone knows how considerate textbook publishers are of their blind customers.

This has inspired several ideas for implementations that would assist with this and similar issues (read as, "it would be really cool if we could:")
* map keypresses from physical hardware such as a bluetooth keyboard to controls, to set up skipping lines and chapters using wasd perhaps, bluetooth keyboards can be accessibly cheap although ymmv for usability due to bluetooth connection quirks, consider other externals such as android wear and unique patterns of omnibutton/play-pause presses.
* enable optional floating softcontrol menu to include long skip (customizable-length multi-sentence skip separate from current short skip), previous chapter/article, next chapter/article, and 'back'.
* utilize a checkbox that will automatically create something like bookmarks for numbered lines on pdf import, to be combined with
* optionally create a link on the small number found in the text with the small instance of the number following, this may not work with OCR and would absolutely require a 'back' or 'return' feature