Grace: One two three four five six seven, eight nine tell and then eleven. Twelve thirteen fourteen fifteen sixteen seventeen eighteen nineteen twenty.
Me: (singing along)... eighteen nineteen twenty...
Grace: No, daddy. You say twen-ty.
Me: Yes, twenty (twenny)
Grace: No. Boys say twen-ty. Girls say twenny.
Me: Grace, everyone says "twenny" most of the time, and "twenty" when they're speaking slowly, or something when you're singing.
Grace: No. Mommy and I say "twenny." You say "twenty." Boys say "twenty"
Me: But Grace, I say "twenny."
Grace: No, daddy. You say "twen-ty". Mommy and I say "twenny."
K (to me): Do we say these words differently?
Grace (insistent): Boys say twen-ty. Girl say twenny.
Where the raw material for this observation came from, I'm not sure. I have sung in many-a-chorus, which training does make me tend to overpronounce my "T"s when singing (I had to unlearn this behavior when singing folky/poppy music where it sounds rather silly to be so hyper-articulated). It's possible that's the root of Grace's observation, though I'm not at all confident that if I could play back recordings of us singing this would be the case.
One of the interesting moments for me in this conversation was when I realized Grace was not being descriptive but prescriptive here. Her initial comment, I'm pretty sure, was motivated by the fact that I had said it like she does (i.e. wrong for a boy), not by the fact that I said it differently from her.
This seems like an interesting window into language change for me. There are a great many variations in language and the human mind is a pattern-hungry and meaning-hungry thing. Obviously the great majority of patterns new speakers pick up (be they sociolinguistic or purely linguistic) are "real" (i.e. observed or obeyed by the vast majority of speakers). But a pattern like this new gender distinction from Grace is the equivalent of a ghost-sighting: she sees cultural significance where there is none. Language change happens the same way mass-ghost-sightings do -- we all have similar minds, see similar patterns, etc. Except that in language there is no "real" outside the minds of the beholders, so if we get enough Graces together, we could produce a generation of women who refuse to pronounce "t"s after "n" for fear they'll sound too masculine (or of women who will start enunciating "t"s to project male power in given scenarios etc etc etc).