Update: Post [l] flapping

During a news story this morning, a young woman spoke of the guilt she sometimes feels regarding any benefit she has received because of her father's death. The setting of the story is West Virginia. Why is the place relevant? Because in her comment the woman (named Amber) used the word "guilty" and pronounced it with the flapped 't' I mentioned in the last post. Evidence then (though not proof) that the environment for flapping has opened up to include [l] before the [t].

To formalize the previous rule we would first assume a post stress position (although that can be complicated/refuted) then we have some choices. I'll present two of them to explain the [t]-->[ɾ] conditioning environment.

We could rely on approximants and formalize the environment as /[+approx][-lat]__[+approx]. This would not require the added analysis of [l] in the pre-[t] environment--so it's not as much fun to do. The [-lat] strikes me as an ad hoc feature requirement.

We could make the argument that [l] is not +continuant before [t], so the rule could be / [+son,+cont]__[+approx]. It would be fun to find evidence in other places that could support the loss of continuance in certain environments.

To re formulate the rule (as it appears some dialects flap in a wider environment) we can change the first environment very simply: [+approx]__[+approx]. This shows enough simplicity for me to accept it. It accounts for the forms I've mentioned. It shows reasonable motivation.

But now we have to complicated things again by introducing a word like 'antifreeze' that shows something very like flapping. According to the last rule mentioned this would be pronounced with the underlying [t]. Why then do so many of us flap or completely elide the [t]? We'll come back to it.