Granular side-by-side diffs, prose inline tracking, Levenshtein edit distance, and vocabulary similarity matrix.