Phrase level highlighting is often too much. Word level highlighting is too little. Allowing the user to determine the length would allow just the right amount of text to appear on the screen, as well as better match pre-recorded audio in the making of audio books.
So, are you thinking that in ever text box you would break it up into what gets highlighted? And then use something fancy (e.g. Aneas) to figure out what audio goes with what text?
Yes, that’s pretty much it, plus such would allow for some shorter highlighted bits, as some sentences are really long (like this one :).