Can't even
« previous post | next post »
In the comments on "Cant-idates" (11/12/2025), there was some back-and-forth about how much phonetic residue Americans generally leave of the word-final /t/ in sequences where can't is immediately followed by a vowel-initial word.
In defense of the answer "not much", I pulled three examples of "can't even" (literally) at random from the NPR podcast corpus I've used in previous posts (and in teaching corpus phonetics).
David Brooks: uh basically, you have uh uh regulators who are asked to reorganize 17 percent of the economy and they can't even do ((it)) a website right
Zeroing in:
Deborah Blum: I- I mean I can't even figure out where they came up with it.
Zeroing in:
John Rother: Each year, the threatened reduction is so unmanageable that we can't even figure out how to pay for it.
Zeroing in:
Quod erat demonstrandum.
I doubt that the /t/ has been fully deleted in these examples, but its realization has been (variously) lenited to the point where the distinction between presence and absence is at best stochastic.
This doesn't mean that the distinction between can and can't is entirely lost, since in a similar context, the vowel of can would generally be reduced.
Don Monroe said,
November 13, 2025 @ 8:50 am
I agree with you about the /t/, although there is a little remnant, but I don't think there's any danger of confusing "CAN'T E-ven" for "can E-ven."
Bob Ladd said,
November 13, 2025 @ 9:41 am
Yes, there are plenty of contexts in which the lenition of /t/ in the environment Vntə is normal in AmEng. This often confuses speakers of Southern BrEng who hear things like "can't even" in informal AmEng and think they've heard "can even", because they'd expect a different vowel in "can't".
But the issue in yesterday's "cantidate" post was whether the distinction between /t/ and /d/ would neutralize, making "cantidate" a good play on words, and I think that commenters (including myself) who said the distinction would generally be maintained were right. That is, Vtə and Vdə both "flap" and tend to neutralize the distinction, but Vntə and Vndə are different, because in that context d -> flap is much less common than t -> flap.
Perhaps Mark can dig up good audio examples, but I'd be surprised to find confusable tokens of e.g. centre and sender or blunter and blunder.
J.W. Brewer said,
November 13, 2025 @ 9:55 am
Even w/o reduction I have a different vowel in (the relevant) "can" than "can't," with unreduced "can" being homophonous with "ken" w/ the DRESS vowel rather than having the TRAP/BATH vowel of "can't." That's apparently a regional variant in AmEng: wiktionary ties it to the greater Philadelphia region (where I was born and raised) although I've never noticed it being one of the stereotypical "tells" that people deliberately mimic when they're trying to affect a Philly accent.
Benjamin E. Orsatti said,
November 13, 2025 @ 1:15 pm
Here's another take — when I (W.Pa.Am.Eng.) speak the words aloud, I notice a "glide" before the vowel in "can't" that isn't present in "can"; something like: "can" –> /k'æn/ :: "can't" –> /kjænʔ/. Anyone else, or is it just something I picked up hanging around Russians?
Victor Mair said,
November 13, 2025 @ 9:44 pm
I've long wanted to ask this question on Language Log, and this is probably about as close as I'll get to have a chance to ask about it. Namely, there's a female vocal group called Wilson Phillips who have a song called "Hold On". Every time I hear them pronounce "Hold On" in that song, I'm struck by how they render it clearly as "Hol-don".
As for "can't", there are quite a few "Cantons" in America. I'm from the one in Stark County Ohio, and the people there pronounce the name not as "Can-ton", but as "Cant-un".