I think having a slider (without any indication of rating/difficulty) is useful in this regard, because it can just be indicitive of the ratio of dialogue to descriptive text. Although having it next to all those other evaluations which do suggest that one side is easier than the other may give that impression anyway… 
Agree! I think that something like “visual storytelling” might also be useful as a complement to this, as the two don’t necessarily have an inverse relationship.
I definitely think they should all be separate and optional - not all measures will apply to all books, and some aspects of language are easier to judge than others. You want users to be encouraged to use the evaluations rather than feeling overwhelmed or confused by them.
Perhaps, since there are lots of ways to judge language, and not all measurements will apply it all book types, they wouldn’t all be visible on the book page at first. There could be a masterlist that you could choose from via a drop-down or search bar on the item page or the additional options pop-up. And once one person has rated a specific value, it is displayed on the book/item page, so you don’t see lots of empty evaluations, only ones that have already been rated by at least one user.
Another way might be to detect specific keywords in a user’s review. For example, if they use the word “grammar”, a suggestion for “Grammar: easy / advanced” would appear, in order to encourage evaluations on relevant aspects of the language.
Some more possible evaluations:
Furigana: all kanji / none
Kanji / kana use: normal / unusual (children’s books written in hiragana, experimental writing, used for specific effects, etc)
Language: modern / old-fashioned
Style: vernacular / literary
All this might also be useful for the ideas about recommendations discussed in the other thread.
Full of ideas, but no clue about coding or tech stuffs… 
