## Metathesis in action

At the end of the May 1 episode of the NPR show "Milk Street", host Christopher Kimball interviews Dr. Aaron Carroll about a recent California court decision that could force coffee to come with a label warning that it contains a chemical known to cause cancer.

The chemical in question is acrylamide, and it's apparently created (in small quantities) whenever carbohydrates are heated above about 250 degrees farenheit — so bread, crackers, cake, cookies, pizza, pretzels, fried potatoes, corn chips, and lots of other things besides coffee that most people eat regularly. Dr. Carroll argues that the quantities of acrylamide involved are far too small to pose any measurable danger, and that warnings like this one have the bad effect of persuading people to ignore all such messages.

But this is Language Log, not Cancer Warning Over-Reach Log, so what's the linguistic point? It's the way that Dr. Carroll pronounces the name of the chemical in question.

The interview starts at about 43:55 in the show — you can listen to the whole seven-and-a-half minutes here, if you want to. The word "acrylamide" occurs five times in the interview:

Each time, instead of /əˈkɹɪləmaɪd/, Dr. Carroll says something like /əˈkɹɪməlaɪd/.

This consistency of /l/-/m/ metathesis means that it's not a speech error, but rather a non-standard lexical entry — an excellent example of non-adjacent metathesis, like that involved in the change from Latin miraculum to Spanish milagro.

1. ### Michael Watts said,

May 2, 2020 @ 3:55 pm

This consistency of /l/-/m/ metathesis means that it's not a speech error, but rather a non-standard lexical entry

While this is possible, I think it's also possible that (1) it's not a word he commonly uses, so he reached for something and went with that for the rest of the interview; or (2) he made a mistake, recognized it, and doubled down on it for the rest of the interview rather than correct himself; or (3) he made a mistake, didn't recognize it, and kept making it for the rest of the conversation without realizing it.

I'd rate (2) as less likely in a vacuum, but they're all possible. I spent a while in a conversation saying "Costa Rica" when I meant "Cuba"; that was not due to a non-standard lexical entry on my part.

2. ### MattF said,

May 2, 2020 @ 4:38 pm

I think that being a polysyllabic technical term that one may read much more often than say makes this sort of error more likely. I had that problem with ‘icosahedron’ for a long time.

3. ### KB said,

May 2, 2020 @ 4:46 pm

Costa Rica exists, "acrymalide" presumably does not.

May 2, 2020 @ 4:46 pm

I actually think Michael Watts's suggestions are less plausible than MYL's original suggestion. If it was a problem of an unfamiliar word, he wouldn't have pronounced it so unhesitatingly and fluently the first time around, and he almost certainly wouldn't have had the metalinguistic awareness to realise he'd made a mistake and keep doing it on purpose to cover it up. And non-adjacent metatheses really are pretty common, both as non-standard lexical entries ("intregal" is a good example, and "nucular" almost counts as a metathesis). For some reason a lot of metatheses that get established historically seem to involve sonorants – there are lots of these in the Romance languages, especially Spanish. (My favourite is Italian coccodrillo for "crocodile".)

5. ### NW said,

May 2, 2020 @ 5:13 pm

If someone is unfamiliar with the word, the familiar word 'acrylic', with chemical connotations, is likely to influence them, surely? And if they have a bit of chemical knowledge but still don't know this word, the ending 'amide' is pretty basic. It's not some randomish string like 'remdesivir' that they have to memorize as a whole.

6. ### Viseguy said,

May 2, 2020 @ 6:10 pm

Alternative theory: Initial error reinforced by repetition (I'm guessing that the interview was probably not the first time the speaker made this mistake). But I like the "milagro" explanation better.

7. ### ktschwarz said,

May 2, 2020 @ 6:28 pm

Aaron Carroll has written a bunch of columns and blog posts about this coffee labeling issue, and here's a video from 2018 where he pronounces it correctly over a dozen times, so I'd say the possibility of non-standard lexical entry is ruled out as well. I think the only remaining possibility is what Viseguy said, initial error reinforced by repetition.

[(myl) Morphophonological metathetic variation! ]

8. ### Bloix said,

May 2, 2020 @ 9:49 pm

The herbicide glyphosate, which is the active ingredient in Roundup, was recently in the news a lot (before we started to have only one thing in the news). I need to write and think about it due to my work. It's hard to remember that it's glyphosate, not glysophate, and mistakes are pretty common. I suspect that the much more common word phosphate, which is similar to both the correct and incorrect versions but is marginally closer to the mistake, interferes with the ability to remember the correct spelling and pronunciation.

May 3, 2020 @ 1:58 am

Bloix's example is great. I completely agree that "glysophate" comes out a lot more naturally than "glyphosate", though the only phonetic explanations I can come up with are pretty much just hand-waving. But there must some kind of notion of "more natural sequence of place of articulation" that affects these, not just influence from other words in the lexicon. An example that I think many English speakers can relate to is kids pronouncing "spaghetti" as "basghetti".

10. ### Peter Taylor said,

May 3, 2020 @ 2:47 am

KB said,

Costa Rica exists, "acrymalide" presumably does not.

I don't know enough to be certain, but a Google Scholar search for acrymalide does turn up a few hits, with particular emphasis on acrymalide gels, and the very relevant acknowledgement

To my two undergraduate students, Peter and Christina (co-founder of Acrymalide [sic] Industries), I must offer unique thanks since their youthful excitement, humor, hardwork, selflessness, snapchats, and consistent efforts have improved both this work and myself.

Badgley, M. A. (2018). The critical role of cysteine import and metabolism in pancreatic cancer (Doctoral dissertation, Columbia University).

11. ### Philip Taylor said,

May 3, 2020 @ 3:08 am

I have never heard a child (or anyone else, for that matter) pronounce "spaghetti" as "basghetti", but as a child I would routinely pronounce "foliage" as "foilage", and "lieutenant" threw me completely. As to glyphosate, I thought that it was "glyphosphate" for several time, and pronounced it as such, but never as "glysophate". For me, these are just examples of "what you think you see is what you pronounce". If children really do pronounce "spaghetti" as "basghetti, is this perhaps because of an increasing exposure to basmatti rice, something that was unknown (in the UK, at least) when I was a child but which is now widely known, sold and eaten.

May 3, 2020 @ 3:49 am

@ Philip Taylor: No, this may be a Brit/Am difference, but it can't possibly have anything to do with basmati rice. I was of the relevant age in North America in the 1950s, and spaghetti was completely normal but basmati rice was unknown, and kids said basghetti anyway.

Also, your example of "foliage" and "foilage" suggests you're talking about words acquired through reading. The evidence for any natural tendency for certain sequences to be subject to metathesis would have to depend on spoken forms.

(And w/r/t "lieutenant", many American readers may be unaware of the presence of /f/ in the British pronunciation, which I presume is the thing that threw you.)

13. ### Kristian said,

May 3, 2020 @ 4:01 am

Was Basmati rice (not basmatti, it's not Italian but Indian) really unknown and uneaten or just not labelled as such?

14. ### Michael Watts said,

May 3, 2020 @ 5:53 am

For "pasghetti", I also never heard the form, but I became aware through American children's literature (Ramona, I believe) that this was a mistake considered typical of children.

15. ### Victor Mair said,

May 3, 2020 @ 6:03 am

"I have never heard a child (or anyone else, for that matter) pronounce 'spaghetti' as 'basghetti'"

In my whole, long life, "basghetti" (or variants like "basgetti", "bisgetti", etc.) is almost the only way I've heard children under about the age of 5 say "spaghetti", regardless of socioeconomic class / standing. For individuals of certain socioeconomic class / standing, this mispronunciation persists into adulthood. In my experience, it is extremely rare for a child under the age of 5 or so to say the word correctly.

It's like Gerald Ford and Jimmy Carter (himself a nuclear scientist) saying "nucular". See especially the comments here:

"Not just more of the same" (9/4/08)

For some historical context, see "'Nucular' solecism traced to 200 B.C.", 1/2/2004; or for a more serious linguistic discussion, "Axe a stupid question", 3/21/2005, and the links contained therein.

If you want to get a visual sense of how widespread "basghetti" is in our culture, feast your eyes on this.

WARNING: If you follow all the links, you can spend at least half a day on the visual culture of "basghetti". E.g., "On top of basghetti, all covered wif cheese, I wost my poor meatball, When somebody sneezed…". There are even multiple representations (in different colors) of Vladislav's condensed canned basghetti ("made with worms") à la Andy Warhol's iconic Campbell's Soup Cans (1962). And don't get (me) started on the gifs, for which I'm always a sucker. Whoops! There I've lost another half an hour of my precious day! And the basghetti Instsagram posts (photos and videos), the witty basghetti themed t-shirts, the cute basghetti themed animals…. Whoops! Whoops! There goes another hour!

… And this is why my eyes are closed
It's just as well for all I've seen
And so it goes, and so it goes
And you're the only one who knows…

16. ### Philip Taylor said,

May 3, 2020 @ 6:15 am

Kristian — I think unknown when I was a child, but certainly obtainable in specialist (Indian) shops ten years later. The nearest equivalent when I was a child was patna rice, and that is what one would use (boiled) if one was going to serve a curry, which in those days was always dumped in the middle of the rice.

17. ### jin defang said,

May 3, 2020 @ 8:00 am

some people find certain words hard to pronounce, so drop a letter or even a syllable—-think about "Febuary" for example. Before the announcements for the DC metro stops were automated, quite a few of the conductors referred to Ju-dish-u-ary Square, the orthodox way being harder to say. And, as someone earlier in the list mentioned "nuke-u-lar."

a P.S.: kids of my acquaintance start out saying p'ssketti, not basketti. The second letter of the standard spelling stays, is not softened to a b.

18. ### Joyce Melton said,

May 3, 2020 @ 8:01 am

I don't think I've ever heard a child or anyone else say basghetti. It always sounds more like pasketty to me, with unaspirated, voiceless stops.

19. ### Philip Taylor said,

May 3, 2020 @ 8:50 am

Are there any similar attested metastheses for "Przewalski", as in "Przewalski's horse" ? It seems that if one can stumble over something as ?simple? as "spaghetti" (/spæ ˈɡet i/), one could certainly stumble over "Przewalski" (/ /pʃe ˈvɑːl ski//), especially if one were not a Polish speaker …

20. ### Victor Mair said,

May 3, 2020 @ 9:09 am

Another of the variants to which I referred.

21. ### Bloix said,

May 3, 2020 @ 9:18 am

Perhaps idiosyncratically, I think of Feb-you-ary as being just as correct as Wenzday. Feb-ru-ary sounds to me like a spelling pronunciation. Like off-ten.

Somewhat related – the Pacific NW bivalve burrowing clam, the geoduck, usually pronounced gooey-duck – the spelling is apparently an absurd mis-transcription of a Native American word.

22. ### Robert Coren said,

May 3, 2020 @ 9:39 am

Mention of being thrown by lieutenant as child reminds me that I encountered the word colonelin writing (identifying a minor character who appears near the beginning of The Voyages of Doctor Dolittle, if I remember correctly) before I ever heard it spoken, or at least before I knew that's how the name of the rank was spelled, and I got it completely wrong. (It didn't help that I somehow got the idea, in contradiction to the testimony of my own eyes, that it was spelled colonol.)

23. ### Robert Coren said,

May 3, 2020 @ 9:41 am

@Bloix: Would you say that Febuary was the last time you could safely go to the libary?

24. ### Philip Taylor said,

May 3, 2020 @ 9:52 am

Robert — In British English, it is not unusual to hear "library" reduced to /ˈlaɪb ri/, at least by those who are not habitually careful in their speech.

25. ### Rose Eneri said,

May 3, 2020 @ 10:14 am

I think the metathesis in this post is due to (IMHO the illogical) stressing of the second syllable of the word. The metathesis would probably never occur if acrylamide were pronounced logically, namely, stress on the first syllable of the first element, acryl, and stress on the first syllable of the second element, amide.

'a cryl 'a mide

Stressing the second syllable of compound technical words often obscures the meaning. When hearing about "hydrogenated" fats, one would have to think twice to realize that these fats have added hydrogen. Why stress the (chemically speaking) nonsense second syllable drodg, especially when doing so also changes the vowel? Similar reasoning applies to numerous words, such as, intercalate.

Can a person who has studied an agglutinative language please tell me if stress is shifted in an agglutinated word to syllables that would not be stress in the individual elements of the word?

26. ### David L said,

May 3, 2020 @ 11:09 am

I don't take the Washington DC metro much anymore, but drivers had a number of pronunciations for the "Judiciary Square" station. Joo-dish-ee-airy, Joo-dish-aree, Joo-dish-a-wary were the most frequent, as best I recall.

May 3, 2020 @ 11:15 am

@ Rose Eneri: The problem isn't the stress shift per se, but the fact that unstressed vowels in English tend to get reduced to schwa. For example, Italian has a completely analogous stress change to English photographer / photographic (Italian fotógrafo / fotográfico (showing the stress with an accent mark, unlike the standard spelling) but it essentially doesn't change the vowels at all, so the connection between the two forms is completely clear, unlike in English. You may well be right that by obscuring the connection to amide, the stress shift in English encourages the metathesis that this post started with.

28. ### Michael Watts said,

May 3, 2020 @ 11:24 am

drivers had a number of pronunciations for the "Judiciary Square" station. Joo-dish-ee-airy, Joo-dish-aree, Joo-dish-a-wary were the most frequent

I would consider the first two of those to be standard pronunciation.

29. ### Rodger C said,

May 3, 2020 @ 12:00 pm

Long ago there was a right-wing-made docudrama about the persecution of Andrei Sakharov (who should have been better served). I stopped watching it after the second or third time I heard Jason Robards, as Sakharov, refer to himself as an "acamedician."

30. ### ktschwarz said,

May 3, 2020 @ 1:43 pm

basghetti vs. pasketty: Great example of how English speakers tend to hear an unvoiced unaspirated stop as voiced. (See Language Log posts on "kiss the sky"/"kiss this guy".) The adults are producing an unaspirated /p/ in sp-, and the kids may be hearing it as /b/ and producing that, or else they may be producing an unaspirated /p/ which the adults hear as /b/.

May 3, 2020 @ 3:03 pm

@ktschwarz – Yep, that's why I spelled it with a B, though I agree with other commenters that can also be P. But some kids acquiring English are definitely unsure of the "correct" phonological status of the stops in sp-, st-, and sk- clusters, and seem to hear them as /b d g/. At least one of my kids tended to spell these clusters as sb-, sd-, sg- when he was first learning to write.

32. ### Bloix said,

May 3, 2020 @ 3:03 pm

Robert Corern – Liberry to my ear is childish. I don't doubt that there are adults who say it but I can't recall it. If I had to bet I'd say Feb-you-ary" more common in daily American speech than Feb-ru-ary.

Ktschwarz – I've always thought of it as pisketti.

Rodger C – I suspect that acamedician arose out of interference from medicine, medicinal, and perhaps magician.

Michael Watts – as a long-time Red Line passenger, my recollection is that most drivers said "Joo-dish-you-ary," but one or two would say, with emphasis, Joo-dish-eeee-ary.

Philip Taylor –
"Przewalski's horse" – for a long time, having never heard it spoken, I thought of it as Prez-valski's horse. I suspect that I read the combination "prze" as prez (i.e. president). But this isn't really metathesis, – it's just unfamiliarity with another language's orthography. In Polish, I now know, rz is a digraph, so nobody says pr-ze-valski.

33. ### Chas Belov said,

May 3, 2020 @ 3:10 pm

@MattF: Ah, yes, I say "isocahedron."

@Bob_Ladd et al: I said "skabetti" for "spaghetti" but I can't remember whether it was initially accidental or intentional. Eventually it was intentional for fun.

34. ### TR said,

May 3, 2020 @ 5:41 pm

So what's the phonological motivation / generalization here? Is it "avoid liquids in onsets of adjacent syllables", "align liquids with beginnings of feet", both, neither?

Rose — lots of languages have accent patterns for compounds that don't necessarily reproduce the accents of the consituent parts, including Latin and Greek (where almost all such English compounds come from).

35. ### ktschwarz said,

May 3, 2020 @ 5:57 pm

Bob Ladd: one of my kids tended to spell these clusters as sb-, sd-, sg- … like the preschooler who wrote sDr for star and SgR for square, discussed at Language Log and All Things Linguistic. These kids are hearing more closely than adults: they haven't yet learned to ignore the distinctions that are not made in spelling.

36. ### Victor Mair said,

May 3, 2020 @ 6:39 pm

See "Dzwil" (11/3/15), in which I described how this is a contraction of the Polish surname "Dzwilewski". I also discussed how this name is pronounced.

The post was followed by 55 comments that touched on many interesting, hard to pronounce names.

37. ### Barbara Phillips Long said,

May 3, 2020 @ 8:43 pm

My eighth grade (U.S.) English teacher was obsessed with making her students pronounce the r in February. She also insisted we pronounce the u in Tuesday — none of that “toosday” stuff for her, and she made a fuss about the pronunciation of “horrible.” (No, she did not make us say “often” with a t.) As a result, we made a lot of lame jokes about “horrible Tuesdays in February” that we found hilarious.

A lot of people I have met don’t pronounce the r in February. When I was young, I considered the r-less pronunciation a class marker. I think I was just enjoying being pedantic; class doesn’t seem to be a consistent correlation. There are plenty of people who pronounce the r, and there are plenty more who don’t.

My former teacher would not be pleased with Tuesday speakers, either. I know of few, if any, who say “tyuz-day.”.

38. ### Andreas Johansson said,

May 4, 2020 @ 12:24 am

@Philip Taylor:

I recently read a book where "Przewalski" was spelled a number of different ways, but most commonly "Prezwalski".

39. ### Philip Taylor said,

May 4, 2020 @ 3:48 am

Andreas — sigh !

Barbara — your 8th-grade school mistress sound very much a woman after my own heart, and very similar to the teachers who guided me in my formative years, but I am intriged to know : for "Tuesday", would she have accepted /tʃuːz deɪ/, or would she have insisted on / ˈtjuːz deɪ/ ? I ask because I know from analysing a long recording of myself that I am not as punctilious about using / ˈtjuːz deɪ/ as I would wish to be, but I still wince when I hear /tʃuːz deɪ/ on BBC radio …

40. ### Robert Ackerman said,

May 4, 2020 @ 3:51 am

Another interesting case in English is the pronunciation for the common root vegetable Brassica rapa rapa, called in most of the English-speaking world "turnip." Except in a part of central England (I've forgotten exactly where, and can't find my notes on sources) and in the Canadian maritime provinces, where it is called "termit." It appears that alveolar nasal sound of "n" has become a bilabial nasal "m", and the bilabial plosive "p" has become the alveolar plosive "t". That is to say that the attributes bilabial and alveolar attributes have shifted from one phoneme to another.

I'm sure someone has a more precise description of the phonemes, and recognizes this phenomenon. I hope.

41. ### Philip Taylor said,

May 4, 2020 @ 4:15 am

Robert — Dartmoor (not too far from my present abode in Cornwall) is one location in which "turnip" is locally known as "turmit" : https://www.legendarydartmoor.co.uk/turni_mang.htm

42. ### Philip Taylor said,

May 4, 2020 @ 4:34 am

(I should have checked the English Dialect Dictionary before posting the above) —

TURMIT, sb and v. In gen. dial. use in Sc. Irel. and Eng. Also in forms 'mip, 'mit Hrf$^1$; tarmit c.An$^1$; tommut Yks.; tormat n.Lan; tormit n.Cy. Nhb$^1$ Shr$^2$; tormut m.Lan$^1$; tummit w.Yks. Lei$^1$; turmat Wm. w.Yks$^1$; termet Sc. Cum$^3$ War. Sus. I.W.$^2$ Wil. Som. Cor.$^2$ turmop w.Yks$^5$; turmot m.Yks$^1$ n.Wil.; turmuck Ant.; turmup w.Yks$^1$ Sur.$^1$; turmut Sc. Glo.$^1$ s.Oxf. Brks.$^1$ Sur.$^1$ I.W.$^1$ Dor. w.Som.$^1$ Dev. Cor.$^1$ [phonetics omitted] sb. A turnip.

May 4, 2020 @ 9:22 am

@ktschwarz: Well, it's not exactly that kids "are hearing more closely than adults". At the level of IPA transcription, the stops in sp- st- sk- clusters in English are "voiceless unaspirated", but (as also regularly discussed on Language Log!) IPA transcription is a blunt instrument. Studies of the local effects of stops on voice pitch (F0) actually show clearly that the stops in sp- st- sk- clusters raise F0 on release, just like aspirated p- t- k- (Helen Hanson in JASA, 2009). The real issue is probably phonological: there is no distinction between p- t- k- and b- d- g- in s+stop clusters, so the child acquiring alphabetic literacy has to make a choice – is that sound "really" a p or "really" a b? My impression is that some kids guess one way and some the other – and to judge from the comments on this thread, some kids who metathesize "spaghetti" go for b and some go for p.

44. ### Robert Coren said,

May 4, 2020 @ 9:57 am

@Barbara Phillips Long: My mother, who although born and bred in New York City, had some British-style speech affectations (some of which I have retained throughout my life, such as /ɑ/ in rather), always said /tjuz dɛı/, and I do at least sometimes. (She also put /ju/ in corduroy and Audubon, which it took me some time to learn that just about nobody else does, at least in the US; decidedly not in illustrated, however.)

She had a friend named Lucy, and it was something of a joke for both of them to pronounce the name as /ljusi/.

As to horrible: this is more a regional thing than a matter of "correctness", surely. /ɑ/ in some areas, /ɔ/ in others. (People who know: am I wrong in thinking that the /ɔ/ version generally goes along with the lack of the marry/merry/Mary distinction?

45. ### Philip Taylor said,

May 4, 2020 @ 12:34 pm

Robert — "am I wrong in thinking that the /ɔ/ version generally goes along with the lack of the marry/merry/Mary distinction?". Possibly in American English, about which I know next to nothing. But in British English, which generally preserves the "marry/merry/Mary" distinction (Wells' TRAP/DRESS/SQUARE), I would say that /ɔ/ is the norm in "horrible", at least in RP/near-RP.

46. ### Ken said,

May 4, 2020 @ 6:19 pm

@Robert Coren (reading "colonel"): I had a similar experience with "sew" as a child. I knew the spoken word but didn't recognize the spelling, so I read it to rhyme with "new".

The horrible thing is that I still do this, and it causes a horrible little brain glitch every time. I hit the word, my inner reader says "soo", and my reading screeches to a halt as I back up and correct it. Do linguists have a name for that?

47. ### Andrew Usher said,

May 4, 2020 @ 7:22 pm

And we just see how IPA can be misleading – Robert Coren and Philip Taylor used the same symbol /ɔ/ for considerably different sounds. The first vowel in 'horrible' is quite different in a standard American and a standard British pronunciation and should not be shown the same. That is why I have always recommended /o/ for the usual American NORTH/FORCE vowel (which nearly matches the British one in a word like 'story') – the difference is that Americans have merged in another vowel ('orr') in a back version of merry = Mary and Sirius = serious.

It could be argued that the American system is more consistent in moving _all_ vowels before /r/ to a special system, whereas the British leaves behind short vowels in open syllables.

For myself, I would actually _expect_ to hear 'acryl 'amide and am pretty confident that the historically correct pronunciation for -amide was the same as the independent word 'amide' (originally /'æmɪd/ and sometimes spelled 'amid').

k_over_hbarc at yahoo.com

48. ### Philip Taylor said,

May 5, 2020 @ 7:27 am

Ken — although not in the same category as your "sew" / "new", you nonetheless remind me that when I first heard the question 那两个人是谁 ？/ 'Nà liǎnggè rén, shì shéi ?' / "Who are those two people", I mis-heard it as 哪里 昂个人是谁 ? / 'Nǎlǐ ánggè rén shì shéi ?' / "<gibberish ?>", and despite the fact that I have now known for some 20 years how it should be parsed, my brain still insists on reverting to the meaningless parsing whenever the phrase flickers through my mind …

49. ### Robert Coren said,

May 5, 2020 @ 9:00 am

@Ken: A Usenet newsgroup (remember those, folks?) that I used to inhabit had a troll for a while who spelled the word for bringing a legal action as "sew", and it became a running gag (still is, among the folks who were regulars on that group).

50. ### Robert Coren said,

May 5, 2020 @ 9:01 am

@Philip Taylor: Yes, I was talking about American English; sorry, I should have been more explicit.

51. ### Michał said,

May 10, 2020 @ 7:45 am

'Przewalski' => 'Prezwalski' plus the existence of the Roland "Prez" Pryzbylewski character in The Wire makes me suspect that changing 'prz?' => 'pr?z' must not be an uncommon thing.