on Jane Friedman site:
There is an abundance of book design software on the market these days. Authors, publishers, and book designers have many options, with each option representing a different set of tools and processes for getting a manuscript ready for production.
But in a publishing industry that is increasingly digital first and born accessible, book design software needs to go beyond perfecting the“look” of a finished product. It has to address a fundamental element that ensures a book can reach the widest audience possible: semantic structure.
Semantic structure is a way of describing the different pieces of your text and their relationship to one another. It’s a digital map of a book that can be interpreted by computers—for example, by ebook reading software or by a web browser. It’s a way of saying “this is a chapter, this is a chapter title, this is a plain text paragraph,” so that the computer knows how to present your text.
You might wonder, “if my chapter titles are bigger than the body text, isn’t it obvious that it’s a title? Is that semantic structure?” Well, the answer is…No! Semantic structure is completely separate from design. Think about a vision-impaired person using screen-reading software to read your book: they can’t see that your titles are bigger than the body text, so relying on design to convey semantic structure has the unintended effect of creating barriers between this user and the meaning of your text.