Sharon Rose reminds me that her 4 1/2-year-old daughter Helen has been pronouncing taps in flapping contexts as [t] for quite some time now. Sharon is Canadian and her parents are English, and due in part to this background Sharon more often than not does not apply flapping herself. Helen’s father Tadesse is from Ethiopia, and there are also very few if any taps to be found in his English.

One might think, then, that Helen has picked up on the tendency in her family’s speech toward the variants without flapping. What’s interesting, though, is that Helen has [t] even where there is no [t] variant; so, words like latter and ladder are both [lætɚ] with a [t].

I have a guess about what’s going on here, consistent with what I’ve said before but probably on as flimsy a limb as what Bob said.

My guess is that Helen has figured out that [t] sometimes alternates with the tap [ɾ] and that [t] is both the basic and the more frequent alternant, even in flapping contexts. She therefore cautiously pronounces [t] much more often, perhaps always, even in flapping contexts. Now consider the fact that [ɾ] and [d] sound much more alike than either sounds like [t], and suppose that Helen therefore can’t yet tell the difference between [ɾ] and [d]. So, even if she hears members of her family say (nonalternating) forms like [lædɚ] with a [d], Helen finds them to be the same as [læɾɚ] with a [ɾ], which she in turn analyzes as /lætɚ/ and pronounces as [lætɚ] with a [t].

This idea differs from Bob’s markedness/underspecification-based idea in that it doesn’t refer at all to the relative markedness/specification of these three sounds, which <a href=”I think has several nontrivial problems. It relies instead on the (I think uncontroversial) acoustic/auditory similarity between [ɾ] and [d] (vs. either of them and [t]), and on the idea (also uncontroversial?) that the relative frequency of allophonic variants may tempt a learner to stick with the more frequent variant.

I’m sure there’s more to say about this, but as I said, I’m out on a limb here. Obviously, Helen’s specific linguistic environment does not necessarily apply to the kids that Sally Thomason talks about, but given the fact that flapping is generally optional and that flapping contexts are plausibly less frequent than non-flapping contexts, their linguistic experiences are probably not too different to matter much.