I haven't read all of the paper (on my To Do list), but it came up in response to a post of mine about how people seem to struggle with grasping the precise meaning of generic sentences about sex or gender.
I'm not feeling very articulate today, so apologies for my laziness but I'll pretty much C&P it here, as I think it's relevant.
---
Language often fails to make statistics transparent.
The following sentences look very similar:
a) Men have penises
b) Men are better at maths than women
c) Men are rapists
But sentence (a) refers to almost 100% of men.
Sentence (b)... A male poster on one thread was defending this or a similar proposition with a graph showing two bell curves which overlapped except for the 0.1% at either extreme.
Supposing this graph were correct, and that it did indeed reflect pure biology not some other biasing factor, that would still leave ability in maths not having any correlation with sex except in some tiny proportion of the population (which blatantly didn't include that particular male poster!).
Sentence (c)... Blokes line up to yell NAMALT about that one. Though not so much about (b).
In (a) "men" = "all men".
In (b) & (c) "men" = "a tiny minority of men".
It's struck me that the language enables someone to say, "I'm a man so I have a penis; I'm a man so I'm better at maths than women," without experiencing any jolt as they shift from one assertion to the other.
Adding NAMALT occasionally doesn't seem to dislodge this mental elision of the two different uses of the word "men" from the discourse.
(And obviously the same is true of the word "women" and of any other word for a group.)