The last two lessons counted words. This one asks what a word is. When someone says a reader needs to know 98% of the words of a page, they have counted in a certain way, and the lab counts in another. The two ways have names, and the difference is about one point.
A word, a flemma, a family
A flemma is a word and the forms made from it by the endings you met in the first courses: -s, -ed, -ing, -er, -est, and the forms that do not follow the rules. Walk and walked are one flemma. A family is a flemma and the words made from it with a beginning or a different kind of ending: happy, unhappy and happiness are one family. The word list counts flemmas. Nation counts families.
Three steps of the lab's count
The lab reads a page in this order. First it asks if a word is on the list, by the rules, and counts it as covered if so. Second, if not, it asks if the word is a name. Third, if not, it takes off one beginning or one ending and asks again. A word that passes the third step is counted as a family word. A word that passes none is not covered.
So the three numbers are in order, each including the one before. On the passage they read 91.0% for the list, 94.3% with names, and 95.6% with families. The second step adds 3.3 points and the third adds 1.3. The passage has 12 family words, each appearing once, and 1.3 points is 12 in 936.
The words a family adds
Set the second menu to the words a family adds. On the passage the rows include delightful, unwelcome, unwilling, recollect and overtaken. On the modern documents the first row is disability, made from dis- and ability, and it appears ten times. The step adds 1.0 point to that text, from 92.7% to 93.7%. On the play it adds 0.8, from 94.6% to 95.4%.
What the step cannot see
The step cuts a spelling and looks at what is left. Notable could be cut to not and -able, since both are on the list; the lab tries note first, so its cut is the right one. In the play, resides becomes re- and sides, and endure becomes end and -ure: there the row is wrong too, and the share counts both. The cut is right for delightful, and the lab cannot tell which is which.
Why the gap is small here
The design notes behind this Subject put the gap between counting flemmas and counting families on general text at three to six points; that is an estimate, not a count you can see. The gap the lab shows on these three texts is between 0.8 and 1.3 points. The lab uses a short list of beginnings and endings, and a full family count uses many more. So a full count would add more than the lab does, though some of the lab's rows are only spellings.
Either way you count, the answer for these texts is the same: below 98%. The words that stand between you and an easy read are the ones in the table of words not covered, and the family step has already taken out the ones it can.