This morning, Stephanie Ruhle, reporting a Michigan school shooting on MSNBC, and confronting the word tragedy (with /ǰ … d/ ), replaced it by tradegy (with /d … ǰ/), transposing the two consonants; she noticed the error, and “corrected” it by, alas, a repetition of tradegy, which she didn’t notice, so she just went on. Then in a later report on the shooting, she again referred to it as a tradegy, this time without noticing.
As an error in spelling — TRADEGY for TRAGEDY — this transposition of consonants is common enough to have been listed in Paul Brians’s Common Errors in English Usage, p. 207 (and on the website), where Brians remarks:
Not only do people often misspell “tragedy” as “tradegy,” they mispronounce it that way too.
Here I think that Brians’s focus on errors in written English has led him astray, led him to treat what is at root an error in pronunciation — with the erroneous pronunciation then carried over to spelling — as an error in spelling that then is then carried over into pronunciation. Admittedly, the latter transfer is part of the story for some speakers, but the problems begin with inadvertent speech errors like Stephanie Ruhle’s. An inadvertent speech error that seems to be part of a larger phenomenon.
Ruhle’s error was an exchange, transposition, or — technical term, metathesis — of two consonants. She didn’t intend to do it, and, at least the first time out, recognized it as an unintended error and tried to correct it. But others might have been less sure, and produced /trǽdǝǰi/ in the belief that it was the correct pronunciation — or an acceptable alternative pronunciation, or even the correct pronunciation for another noun somewhat similar in meaning to /trǽǰǝdi/ but distinct from it.
Then /trǽdǝǰi/ can spread to other speakers (who don’t know its history), and in any case some of these speakers will spell the word TRADEGY, because that’s the way it’s pronounced. And then we can get things like this (from an Amazon.com page):
Note that the titling of the book and the design of the cover were done by one set of people and the blurb from the publisher by another, and they’ve ended up differing from one another. And that you could also get to TRADEGY by a transposition in writing the letters, as some dyslexics are inclined to do. There are several alternative routes, all possible, to the text you see here.
Metathesis of consonants. As a speech error. Even just within words, there are several types, but a very common type is metathesis of consonants in parallel positions — especially where both are syllable onsets, as in
Kimura > Mikura (onsets k… m… > m… k…)
(This example and the others that follow in this section are from section M (within–word errors) of the appendix to V. Fromkin’s Speech Errors as Linguistic Evidence (1973).)
In the Kimura example, the consonants involved — m and k — are not very similar phonetically, but there seems to be some tendency for metathesis to be favored for consonants that are highly similar, in particular the nasal stops articulated towards the front of the oral tract, m and n (which are acoustically very similar), and the voiceless oral stops articulated at the periphery of the oral tract, p and k (also very similar acoustically, despite the physical distance between their points of articulation). Examples from Fromkin for the anterior nasals:
dynamics > dymanics (with onsets n… m… > m… n…)
terminus > ternimus (with onsets m… n… > n… m…)
and for the peripheral voiceless stops:
caterpillar > patterkiller (with onsets k… p… > p… k…)
pancakes > canpakes (with onsets p… k… > k… p…)
To which I now add the example we started with, the medial voiced obstruents:
tragedy > tradegy (with onsets ǰ… d… > d… ǰ…)
plus prodigy > progidy (with onsets d… ǰ… > ǰ… d…) [also well attested]
Order effects. Somewhat surprisingly, there is some evidence that certain orders of consonants are preferred over others, so that one kind of transposition is favored over the other, in particular for the sequences involving m and n. From my 1/10/18 posting “Puns and metatheses”, about, among other things:
a tendency [note: tendency], in child language and more generally, to favor sequences of stops ordered from more front to more back points of articulation. In particular, as reported in A. M. Zwicky & A. D. Zwicky, “Patterns first, exceptions later” (in Channon & Shockey (eds.), To Honor Ilse Lehiste: Ilse Lehiste Pühendusteos, 1987):
A number of children acquiring English have been reported as sporadically or even regularly replacing n … m sequences by m … n; our daughter Elizabeth, who for some months replaced animal by aminal and cinnamon by cimmanon, indeed transformed the first of these sequences in novel or nonsense words into the second.
As far as I know, no one’s looked at kids to see if they’re preferentially inclined to replace ǰ … d by d … ǰ rather than the reverse. (Not easy experiments to design.)
December 4, 2021 at 8:09 am |
As I recall, my brother and I had trouble with stamina, which tended to become stanima, probably under the influence of the more familiar animal.
December 5, 2021 at 9:14 am |
First, tendencies will have counterexamples; the question is what the preponderance of examples shows. And often counterexamples have an account of their own — as with the model of “animal” here.
December 4, 2021 at 2:42 pm |
I have heard many people say Calvary for cavalry. Some are aware they do this, and if they know that the context will call for it will disclaim it in advance.
December 4, 2021 at 11:56 pm |
I have the feeling that “jewlery” and “nucular” (and maybe “calvary”) are a different, though possibly related, phenomenon.
December 5, 2021 at 9:10 am |
Yes, these look like reorderings to improve syllable structure. On the other hand, “revelant” for “relevant” looks like an exchange designed to achieve front-to-back ordering of consonants in successive syllables.
December 5, 2021 at 1:31 am |
Surely it’s relevant that, unlike the other metatheses, “Calvary” is a real word, and a familiar one to many.
December 5, 2021 at 8:22 am |
This reminds me that we also had a tendency to say “revelant” for “relevant”.
December 5, 2021 at 8:52 am
But never “ephelant” for “elephant”. Sorry, I seem to have Chico Marx echoing in my head.
December 5, 2021 at 9:03 am
To J B Levin: actually, “efelant” for “elephant” is such a common kid error that there’s a Disney efelant (a fantasy elephant-like creature) character (originally in French).