Ah! I see! So it is a non-obvious (to me) statistical phenomenon, after all. Yes, true, it is much more likely for the known words to be repeated across lessons than the new words. Now it makes sense to me. Repeated known words count for every individual lesson anew, but for the whole course they count only once.
Let me create a hypothetical example to help myself grasp this. Let’s say bold are “new words”.
- Lesson 1: “The man gave me a book.”
3 known words, 3 new words - 50% - Lesson 2: “The woman gifted me a pen.”
3 known words, 3 new words - 50% - Lesson 3: “The teacher handed me a pencil.”
3 known words, 3 new words - 50% - Lesson 4: “The boy sent me a message.”
3 known words, 3 new words - 50%
Overall:
- 3 known words: The, me, a
- 12 unknown words: man, gave, book, woman, gifted, pen, teacher, handed, pencil, boy, sent, message
80% new words
Alright, well, that does clear things up. Thank you!