Today’s NY Times Magazine carries Gabrielle Hamilton’s piece on what sounds like a delicious condiment:

For the entirety of Prune’s 20 years, I’ve confined myself — with pretty strict discipline — to cooking within a European-and-Mediterranean idiom. […] But when I’m cooking at home, there have never been any equatorial or hemispheric confines — mapo tofu, larb salad, goma-ae, smorrebrod all in rotation, with free abandon. And recently, a heaping spoonful of zhug pretty much every day. With this bright green Yemeni sauce, spicy from serranos, perfumy from cilantro, I’m getting ready to expand the territory at work.

I’ve had both the green and the red versions at Yemen Cafe in Hamtramck, a suburb of Detroit that used to be heavily Polish but now has immense Yemeni and Bangladeshi populations. During a recent visit this summer, I stopped in to finally find out how to pronounce “zhug,” thoroughly annoyed by my own chronic stumbling over the “zh” as if it were the leg of a chair in my own den that I habitually stub a toe on. […]

I asked the guys at the cafe to repeat “zhug” for me over and over again to be sure I got it, because the sound was so wildly unlike the English transliteration. I’d been walking around using soft z and a hard g and was so incredulous at their insistent hard s and the hard k at the ending that finally one of the teenagers at the register wrote it down phonetically on an order pad: s-a-h-a-w-k. He pronounced the “haw” portion exactly like the hao in Mandarin ni hao, and very kindly put an extra-large container of the stuff in my takeout order. I sat outside in the privacy of my parked car for a few minutes practicing the pronunciation and taking little sips from the container he had packed.

There are several points of language-related interest here; I imagine a lot of my readers would never guess that Hamtramck is pronounced /hæmˈtræmɪk/ (ham-TRAM-ik), and I personally find it odd to spell “mapo tofu” with only the first half italicized, though I understand it (“tofu” is an English word, and if the whole thing were in romanized Chinese it would be mapo doufu). But I’m posting because of the word “zhug,” which is the stupidest spelling I’ve seen since “geoduck,” pronounced /ˈɡuːiˌdʌk/ (“gooey duck”). I cannot mentally pronounce it any other way than /ʒug/, and I imagine most English speakers would try /zʌg/; /saˈhawk/ is simply not a possible reading. It seems the spelling is from the Hebrew form (Hebrew: סחוג), which Wikipedia renders zhug, zhoug, zkhug, or s’hoog (the last being the best in terms of suggesting how it’s pronounced); the guys at the cafe were using the Arabic sahawiq (سَحاوِق). I’m guessing it’s from the Arabic root سحق ‘crush,’ but of course if anyone knows more I’d like to hear it. In any case, what an orthographic nightmare!


  1. Looking at the Hebrew version of the Wikipedia entry for zhug, apparently both zxwk and sxwg are accepted. Blame it on the fact that we use consonant+h for all sorts of things (th, sh, kh, zh) that would be better served with a letter of their own. Θeir own, I meant to say 😀

  2. Definitely pronounced /sxug/ in common Israeli Hebrew. It’s basically the go-to lingual “spicy thing”, equivalent to Tabasco sauce in WB cartoons.

  3. Skhug is delicious. It’s made from peppers, coriander seeds, cumin, and a few other things.

    From what I understand, in Yemeni Arabic it’s pronounced [zħuːg] (not in the south, which has a q). In Standard Arabic it would be [sħuːq]. In Modern Hebrew it’s [sxug] or [sħug], depending on your dialect. In some parts of Yemen it’s called bisbās.

    It does come from the ‘crush to a powder’ root. In other words, it’s pesto.

  4. The “go-to spicy thing”, if I may contradict my learned colleague, is ħarif, ‘spicy’. That’s the red stuff that you ask to put on your falafel. It may be skhug in its composition.

  5. Stu Clayton says

    The “go-to spicy thing” … is ħarif, ‘spicy’. That’s the red stuff that you ask to put on your falafel.

    In terms of etymology and composition, this ħarif must be related to what in Germany is available everywhere as harissa. The WiPe says”هريسة‎ harīsa, from Maghrebi Arabic”.

    I never ask to have it put on my falafel, because the taste of the canned stuff they use is spoiled by vinegar or something else acidic – which presumably was added to keep it from spoiling. Just like that revolting “sambal oelek” in little glass jars. And don’t get me started about jalapeños en escabeche manufactured here, which are by definition preserved in vinegar and yet they get that wrong too.

    Harissa is the run-from spicy thing in Germany.

  6. ħarif must be related to what in Germany is available everywhere as harissa. The WiPe says”هريسة‎ harīsa, from Maghrebi Arabic”.

    Seems hardly likely. The triconsonantal roots have only an R in common.

    BTW, is ħ the same as ḥ? Is ħarif related to ḥarf ‘letter, character’?

  7. Stu Clayton says

    Well, to the ignorant eye they seem related. Is that better ? “They could be ignorantly related”, I should have written.

  8. PlasticPaddy says

    German herb is apparently not related to Latin herba and is not given a PIE etymology. Is this one of those words which “suggests” a Phoenician substrate for Germanic?

  9. Stu Clayton says

    To my apparently ignorant mind it might suggest that, yes, provided I knew anything about Phoenicians. But why do you hedge with “apparently not related” ? I don’t expect any cautious concessions, as if I secretly might know more than I let on.

  10. <the taste of the canned stuff they use is spoiled by vinegar

    Aargh, just like Tabasco sauce all right.

  11. Isn’t “geoduck” a fairly recent deformation of “goeduck”? The latter makes a bit more sense. (“Fairly” recent in that a 71-year-old remembers “goeduck.”)

  12. Isn’t “geoduck” a fairly recent deformation of “goeduck”?

    According to the OED (Third Edition, March 2012), they both go all the way back:

    α. 18– geoduck, 19– geoduc, 19– geoduk.
    β. 18– goeduck, 19– goeyduc, 19– goeyduck, 19– gooeduck, 19– gooeyduc, 19– gooeyduck, 19– gooiduck, 19– gueduc, 19– gweduc, 19– gweduck, 19– gwiduck, 20– gooeyduk, 20– gooyduck.

    Etymology: < Lushootseed (Puget Salish) gʷídəq < a first element of uncertain origin + əq genitals, apparently on account of the shape of the mollusc.
    The forms with geo– (α. forms) probably originated as editorial alterations of forms in goe– (although these are first attested slightly later; compare β. forms), perhaps influenced by geo– comb. form.

    First citation:

    1881 H. Hemphill Let. in Bull. U.S. Fish Comm. 1 21 Glycimeris generosa. Olympia, Washington Territory… The boys at Olympia call them ‘Geoducks’; they dig them on a certain sand bar at extreme low tide, and sell them to a merchant who ships them to Portland, Oreg… The boys inform me that the Indians on the Sound call them Quenux, and dry them for food with the other clams.

  13. PlasticPaddy says

    For Latin herba I have seen a derivation from *bhar (which would make it a cousin of Latin far) and a derivation from *ghreH1 “to grow, become green”. For German herb there is no suggestion it was borrowed and wiktionary says it is from Proto-Germanic *harwaz. DWDS says it might be related to PIE *(s)Ker “to cut” (compare Schere). This would also connect it to Herbst, which would be nice☺. The Semitic root hrf has a tri-consonantal (or at least a bit better than with harissa) and semantic match with the Proto-Germanic form, but there is no other reason to think there is any relation.

  14. Stu Clayton says


    Only now do I understand what you mean by “German herb” in connection with Latin herba. I had thought you meant (Küchen)kraut (Eng “herb”) and couldn’t remember the German word, thus the circumlocution “German herb”.

    No, I would have not surmised any connection between Ger. herb and Lat. herba, or herpes either. Chipped and worn as they are, all my cups were still there when last in the cupboard I looked.

  15. Anyone who would transliterate Hebrew סחוג to zhug is incompetent in two languages. Better would be s’houg or s’choug.
    Anyway, in my experience Israelis generally call it harif (hot sauce).
    I have given up pronouncing hummus correctly when I go to a restaurant. It’s huhm-muhs.

  16. I’m guessing it’s from the Arabic root سحق ‘crush’


    Stu Clayton says:
    In terms of etymology and composition, this ħarif must be related to what in Germany is available everywhere as harissa.

    Nothing to do with one another. Hebrew ħarif is from the root ħ-r-p which has to do with picking, plucking, being sharp (thus the sharp taste of something spicy). Harissa is from the Arabic root h-r-s which has to do with crushing, mashing (indicating the manner in which harissa is produced). All they have in common is an R in the middle of the root.

    juha says:
    BTW, is ħ the same as ḥ? Is ħarif related to ḥarf ‘letter, character’?

    Yes, two ways of transliterating the same letter, ح in Arabic or ח in Hebrew. It seems like the two might indeed be etymologically related; see here.

    Bloix says:
    Anyway, in my experience Israelis generally call it harif (hot sauce).

    Zhug is (a kind of) harif, but harif is not zhug. If you ask for harif in a restaurant, generally you’re not going to get zhug, just as in the US pasta salad may be a type of salad, but if you ask for “salad” without qualifiers, you’re going to get something very different.

  17. I think writing that sound ”ḥ” is a lot better. Seeing something that starts with h-bar (especially italicized) primes me to expect a very different kind of expression.

  18. Anyone know which Lao/Thai transliteration scheme got us “larb” for what is pronounced [lâːp]? A google around hasn’t yielded much in the way of the world’s history (but I will admit, I’m not super good at this). I thought it might be some sort of transliteration like Gwoyeu Romatzyh or tones in Hmong, but nothing I’ve found points to there being an old scheme that represented Lao tones with another letter.

  19. Goeduck: That’s interesting, I was vaguely thinking it was Algonkian. Got my coasts mixed up. Probably thinking of quahog.

  20. Anyone know which Lao/Thai transliteration scheme got us “larb” for what is pronounced [lâːp]?

    Looks to me like your basic nonrhotic (and evil) use of -r to indicate vowel length/quality, but that’s just a guess.

  21. If you like geoduck you’ll love Betelgeuse.

  22. Since I pronounce that “Beetlejuice,” it doesn’t seem that weird to me.

  23. Charles Perry says

    In a lot of places z’hug could be described as a crushed spice mixture containing fenugreek (the Yemenis are the biggest fenugreek-eaters in the world, it contributes measurably to their protein intake), but in San’a’ sahawig is like a Mexican salsa cruda: tomatoes, garlic, herbs and spices, finely crushed with a stone.

  24. Mmm! I want some!

  25. It’s at times like this that I think of the Descendents’ power-packed 15-second “I Like Food.” (Somewhere I may still have the “Fat” EP.)

  26. Good lord, and apparently the band is still around! (For values of “still” that include years-long hiatuses.) Bratty punk doesn’t often display that kind of longevity.

  27. “In July 2016, Milo announced he would be leaving his scientific career to pursue the Descendents full-time, citing burnout with biochemistry and getting laid off from DuPont.”

  28. John Cowan says

    BTW, is ħ the same as ḥ?

    Yes, they are both the voiceless pharyngeal/laryngeal fricative. The first is IPA and Maltese orthography, the second is Arabic/Semitic transliteration and transcription.

  29. David Marjanović says

    I imagine a lot of my readers would never guess that Hamtramck is pronounced /hæmˈtræmɪk/ (ham-TRAM-ik)

    I didn’t before you blogged about it sometime within the last, uh, few years. 🙂

    I think writing that sound ”ḥ” is a lot better. Seeing something that starts with h-bar (especially italicized) primes me to expect a very different kind of expression.

    ħ is the actual Maltese letter for that sound, and the IPA uses it because it doesn’t like diacritics for places of articulation.

    But interpret it as Cyrillic, and suddenly it’s at the end of my name…!

    This would also connect it to Herbst

    But that’s the cognate of harvest, and while rw > rb is a thing in German, rw > rv is not one in English… given its meaning (“somewhat bitter”, “not sweet or acidic”, mostly of wine and beer), I’ve always taken for granted it’s borrowed.

  30. I misplaced my copy of the Language Hat style manual, and so I don’t know what it says about ħ/ḥ.

    (Nerd tip: if you use a Mac, with the “ABC-Extended” keyboard, ħ id option-l + h, ḥ is option-x + h. Or you can install IPA Palette.)

  31. “I cannot mentally pronounce it any other way than /ʒug/.”

    My mental pronunciation of the unfamiliar word in this post’s title was /ʒʊg/

    /ʒug/ would be “zhoog” (unless I thought it was French “jougue”).

    OTOH “wug” is /wʌg/ ; the vowel proves its Anglo-Saxon origin.

  32. nonrhotic (and evil)

    Hat, you’re probably right. That is a spelling choice that I have shifted to the far corners of my memory.

  33. J.W. Brewer says

    Re “that kind of longevity,” quite a lot of vintage punk-rockers who have survived the ravages of the actuarial tables are back out on the road these days. Just yesterday evening I went over to Brooklyn and saw Stiff Little Fingers (first record released 1978, frontman almost five years older than Dr. Milo), with the Avengers (first record released 1977, frontwoman a bit over four years older than Dr. Milo) as the opening act. Before launching into “Teenage Rebel,” which she first sang on stage when she was 18 or 19 herself, Ms. Houston (now aged 60) asked rather plaintively if there were any actual current teenagers in the largely middle-aged crowd.

  34. Stiff Little Fingers — what a band! “Alternative Ulster” is one of the all-time great songs. I’m jealous you got to see them.

  35. PlasticPaddy says

    The teenagers are scraping together the money to hear artists of their generation ( who probably do not speak to us). If they fail, they can look forward to the far future, when they will have more money and these artists’ gigs will cost less. I think these teenagers will also then be able to concentrate on the music rather than contend with rampant hormones.

  36. I have to admit the last current song that really grabbed me, in that good old “I have to buy this immediately” way, was “Uptown Funk,” which of course was a throwback to the ’80s.

  37. From my own teenage years, on down to my daughter’s, there has always been continued interest in listening to 1960s stuff. Material from the 1970s and 1980s is heard much less, although some stuff I thought I would never have to hear again in 1992 has made a bit of a comeback. The disco and punk and prog rock that was around when we were quite young was not something that teens of my generation showed much interest in, once we were making our own music selections. The 1960s, though, has a huge weight of Baby Boomer nostalgia, which has suffused our culture for far longer now than the 1960s actually lasted*. Lillian likes “The Times, They Are A-Changing,” but she does not appreciate it if I point out that, “Your sons and your daughters are beyond your command,” was written referring to her grandparents.

    * Compare the Wild West, which lasted maybe thirty years, but has been romanticized for well over a hundred now.

  38. J.W. Brewer says

    SLF played in Boston two nights before they were in Brooklyn but I’m a bit vague exactly how far west in Mass. chez Hat is located so maybe that would not have been that much better. Although it looks like the Flamin’ Groovies are playing tonight at the Iron Horse in Northampton and if you’ve never seen them you really ought to.

  39. My god, another blast from the past! I used to play the grooves right off Shake Some Action. Alas, it would take even more than the Flamin’ Groovies to pry me out of my hidey-hole and keep me up past my bedtime. Truly I have become an old fart.

  40. J.W. Brewer says

    Well as I noted there are also those who are permanently retired from live performance due to the regrettable but remorseless workings of the actuarial tables, like this avant-garde celebrator of old-fartness: https://www.youtube.com/watch?v=ACl4ZZRJe48

  41. I love that they include “oh man, that’s so heavy” on the live performance. All those little oddities that seem accidental were always part of the Captain’s plan.

  42. Captain Beefheart is another old fave! I’ll never forget my pal Kirk (a loyal son of Detroit, who swore by the MC5) introducing me to Trout Mask Replica one night in 1981 (when NYC still had Checker cabs — eheu fugaces…).

  43. J.W. Brewer says

    Never got to see the Cap’n. There’s at least one Magic Band alum who plays around NYC with some frequency (Gary Lucas) but I haven’t managed to catch up with him yet. I saw a surviving quorum (the three then-living members of the original quintet) of the MC5 play a free show in Central Park in maybe ’06 or ’07 and then saw the golden-anniversary-cash-in “MC50” (Brother Wayne Kramer plus four younger guys closer to my own age who were happy to play the roles required) a year or two ago at Irving Plaza. I feel like Hat needs to get out of my cool obscurantist record collection and listen to that more soothing aging boomer stuff like James Taylor and badly reviewed albums by ad hoc combinations of some but not all of CSN&Y. Or should I go the other direction, mention I recently saw Fred Frith, and see if Hat has a bunch of Henry Cow anecdotes he was just waiting for an opening to share?

  44. Never got into Henry Cow, but I’m a Pere Ubu fan, if that helps.

  45. Also Essential Logic; after I discovered Lora Logic (via X-Ray Spex, of course), I tried to collect every piece of vinyl she’d ever recorded. Imagine my chagrin when she became a Hare Krishna!

  46. (No offense to Mr. George ‘arrison.)

  47. J.W. Brewer says

    I can’t speak for Brett’s teenage daughter and old hippie stuff, but the older of my two teenage daughters once went to school (I think she was still in 9th grade so this would have been 3.5 years ago) wearing a late ’80’s Pere Ubu t-shirt she had fished out of the back of my closet, from the period when they were using “The Return of the Avant-Garage” as their catchphrase. I think she just thought the shirt looked cool compared to old-band shirts of the more common Pink Floyd / Guns N Roses genre other kids had gotten at the mall. She turned out to have no interest in actually listening to the band’s music, and put on a classic 15-year-old flat/impassive affect when I firmly insisted on playing “Final Solution” at high volume on the car stereo while driving her somewhere.

  48. Lora Logic! My one measly connection to the punk rock scene is that my older brother played bass on their first EP. But then he was replaced and that was pretty much the end of his musical career.

  49. He’s a hero to me!

  50. > nonrhotic (and evil)

    I can confirm (as someone currently living in Thailand) that non-rhotic spellings of this kind (i.e. “larb” for /la:p/ ) are extremely common here. As a non-rhotic speaker myself, I would of course characterize these kinds of spellings as “perfectly natural” rather than “evil”, but there you go.

  51. *makes apotropaic gesture*

  52. Like sex, popular music, racing cars and sex, of course everything non-rhotic is evil. That’s why it’s more popular. Now let’s just move on.

  53. The teenagers are scraping together the money to hear artists of their generation

    They’re probably also saving quite a bit from being able to sample everything on Spotify and/or Bandcamp, though, instead of having to fork over for entire LPs on the basis of maybe two songs on the radio. Everything reasonably current and/or commercialized at least, which should be enough for most teenagers. (Those few who are already into the Henries Cow and Peres Ubu might today have a worse time than us 90s teenagers did with P2P, but then that’s not really potatoes to potatoes anyway…)

    Unrelatedly, after re-reading the entire spelling subthread, I submit that leopard is now to be pronounced as /luːiːpɑd/ (rhyming with “gooey cod”).

  54. Works for me. Let’s add to the weirdness of English!

  55. That will never work. Guepardo is cheetah.

  56. John Cowan says

    of course everything non-rhotic is evil.

    By no means. It’s respellings that work only for the non-rhotic that are evil.

    That’s why it’s more popular.

    Maybe, maybe not. Counting only L1 speakers, rounding to the nearest million and omitting countries with less than a million, I make it 225 million non-rhotics (England, Wales, Australia, N.Z., South Africa, Southern U.S., African Americans not in the South) and 225 [sic] million rhotics (Scotland, Northern Ireland, the republic of Ireland, the rest of the U.S., anglophone Canadians). That’s very rough, of course, and uses first-impression figures that are probably from slightly different years.

  57. J.W. Brewer says

    To J Pystynen’s point, I just pulled up Spotify on my computer at work and it appears to lack any of the canonical Henry Cow studio albums from the Seventies but does have them doing a John Peel session from 1973, because Spotify must have gotten the rights to the entire John-Peel-Session-corpus in one fell swoop without distinguishing between more and less obscure artists therein. So I’m listening to “Nine Funerals of the Citizen King” right now.

    EDITED TO ADD: OK, Spotify has (as part of a larger thing that must have been a CD reissue many years after the fact) the 1975 “Desperate Straights” album Henry Cow did in collaboration with Slapp Happy, but coded with credit only to Slapp Happy as the artist. Bad metadata strikes again.

  58. John Cowan says

    Leppard in Regularized Inglish. Leopard would indeed have three syllables, lee-oh-pard. “When you see a wurd, you knoe how to pronounce it.”

    I have propósed an àncíllary convéntion for díctionàries, some chíldren’s books, and téxtbòoks for fórreners: in múltì-syllábic wurds, mark stressed vówels with acúte, ùnstréssed but ùnredúced (aka “secondárily stressed”) with grave. All úther vówels, whéther schwa/schwi or sílent or just the sécond vówel in díphthongs, ar ùnmárked. Some péeple wood próbably write díctionaries instéd, though prólly, próbly, and díctionries wood still count as mìsspéllings. Note that iámbìc rèvérsal and contrástiv stress is not dìfferéntially marked, thus ùnstréssed and not únstrèssed above. (Hoo, that wasn’t éasy!)

  59. Oh come on. We’re a tiny bit evil. Actually no one even realises this is a problem for Americans. I only even noticed it when it was pointed out at Language Hat. For one thing, there’s rhotic and rhotic: what you’re talking about is the big rolled ‘Merican Rs someone like John Wayne uses, whereas the Scots for instance have a trilled R that’s more like Bergensk than American. Neither the Scots nor Norwegians reading English complain about English R-spelling being misleading. People in Boston and Cape Cod speak quite normally despite having had more recent Irish immigration, which leads me to guess that accents in say the midwest and other areas that developed later and much quicker might be less stable. So maybe California will develop an accent with overtones of Mexico and the Pacific Wim. Once Hollywood cottons on, there’ll be no pwoblem.

    The John Peel Henry Cow is also available on Youtube: https://www.youtube.com/watch?v=2NJ27tjPNgo

  60. David Marjanović says

    Neither the Scots nor Norwegians reading English complain about English R-spelling being misleading.

    That’s because they don’t notice, and that’s the detail the devil is in.

    Consider Burma. In nonrhotic (!) German, Myanmar is not a problem, but the two spellings Burma and Birma get two different diphthongs instead of the same monophthong. Then talk about la Birmanie in rhotic French, and all is lost.

  61. Aha, gotcha. All the rhotic nuances are different, you can’t please everybody – that is really quite evil. Burmah Oil was a big company that seems to have stuck their H after the wrong vowel for the rhoticly challenged. I can’t find too many rhotic Rs in Scottish accents, just Edinburgh and a couple more really. The American rolling R seems most similar to N. Ireland accents. You’d think they might complain. I spent a long time searching for the vanishing Northumberland Burr (Northumberland & Geordie may be my favourite English) but none of the sound clips seems to work.


  62. David Eddyshaw says

    I speak a sort of rhotic RP, but am only annoyed by the Presumption of the Non-Rhotics when it comes to UK cryptic crosswords, which all seem to have adopted the convention that non-rhotic is standard. The undoubted Eviltude of this will only be apparent to other devotees of Real Crosswords. Take my word for it. Evil.

  63. the big rolled ‘Merican Rs someone like John Wayne uses

    I poked around on YouTube and found this clip from the 1963 Wayne vehicleMcLintock. At 1:33 he says “But, pilgrim, you caused a lotta trouble this morning, mighta got somebody killed.” Now trouble has an /r/ that we both have, and morning has an /r/ that only rhotics have, and damned if they don’t sound exactly the same to me. What is more, they are the same as the pronunciation of trouble by the announcer of this trailer for the 1970 Leslie Phillips vehicle Doctor in Trouble at 0:10 — not the whole word, mind you, just the initial tr-.

    Scots for instance have a trilled R that’s more like Bergensk

    Up to a point, Minister. Few Scots use that any more. For the most part they use the same /r/ as English and Americans, or at most the tapped /r/ of Spanish r rather than rr.

    Not that these sounds were unheard-of in England as little as a hundred years ago. Here’s John Gielgud in the 1920s delivering Othello’s speech beginning “My most potent, grave, and reverent signiors / My very noble and approved good masters”. Things are very very different! Grave has the same old standard form of /r/ you’d expect, but reverent is a full trill (Bergensk, or else Spanish rr) on the first /r/ and a tap on the second. What is more, the /r/ in signiors is American, and so is the long o in noble and potent. All of which reflects much older forms of RP-ish speech, not Gielgud’s natural accent at all.

    Neither the Scots nor Norwegians reading English complain about English R-spelling being misleading.

    They’re much more used to the convention than we are.

    People in Boston and Cape Cod speak quite normally

    We call that area New England for a reason (but see below).

    accents in say the midwest and other areas that developed later and much quicker might be less stable

    Non-rhoticity is on the retreat everywhere in America: in the cities of the Eastern Seaboard and the Gulf Coast, including NYC and New Orleans, but always excepting Philadelphia, which never had it; in Eastern New England, where it used to extend clear to the Connecticut River, but has retreated far to the east and somewhat to the north, now excluding Rhode Island; in most of the South; in the tiny bits of Canada, such as Lunenberg, Nova Scotia, that ever had it. (In Dublin too, for that matter.)

    Only in African American speech is it thriving; indeed, that variety is more non-rhotic than the Queen, saying “Mistuh Adams” rather than “Miste-r-Adams”. The association with African Americans is one reason non-rhotic American speech is now decidedly non-prestige.

    A particular feature of rising and falling rhoticity is that the NURSE vowel is the first to go rhotic and the last to go non-rhotic. My non-rhotic upstairs neighbor is nevertheless rhotic in NURSE words.

    the vanishing Northumberland Burr

    It’s the Norwegian skarre-r, used in southern and southwestern Norway, shown in green on this map.

  64. the convention that non-rhotic is standard
    If you’re including the Guardian, it must be a breach of its bylaws. I’d send Katharine Viner a solicitor’s letter (surgeon’s letter too intimidating).

    The Guardian, whose headline today is:
    Glastonbury / Queen won’t play festival without badger cull apology

    She’s 93, you know. Most people would have retired.

  65. Lars Mathiesen says

    In the mail I got at 0800 UTC, it said “Brian May: Queen won’t play Glastonbury without badger cull apology”. Sort of harder to misread. Also it was in the ‘Most Read’ section, the front page headline was about the Vietnamese trafficking victims.

    Brian May is 72, anyway. Same age as Lizzie, more or less. More if you account for lifestyle, I should think.

  66. They changed it (it happens all day long). You clearly need to roll out of bed a tad earlier, Lars Mathiesen.

  67. Trond Engen says

    Skarre-r and Bergensk r are the same thing, uvular r. There are local differences within the area in details such as degree of, eh, postuvular nonrhoticity, with Bergen maybe more towards the non-rhotic side. The old Scottish r is more like the anteuvular western r which still dominates from Sogn to Sunnmøre.

  68. Yeah, Bergen never sounds very rhotic to me, but I’m not certain I really understand what rhotic is. Up until now I’d been thinking of it only as American R.

    Listen, FAR more interesting in the Guardian than the Queen headline is a really useful 2015 article I just found about French swearing in the TV series EngrenagesSpiral as it’s known on the BBC. It’s my favourite TV detective series of all time, now in its seventh series. I do hope there’s a way for Americans & others to see it, it gives an interesting glimpse of the French legal system amongst other things. This piece is written by Alison Crutchley who seems to know something about language although there’s some doubt in the comments about the gender of putain.


  69. Lars Mathiesen says

    I sort of assume that the version in the mail is the first edition. They can’t change the text in my inbox retroactively. I hope. Also the frontpage image was from the BBC, but they changed their article since then… Try this one at the Newsmuseum, it’s explicitly marked Edition 1 and Brian May is not in it. (I think it’s a print-ready file for the US printers, so the 19:10 timestamp may be any time zone from BST to PDT).

    But to be sure, it often happens that following a link from the mail brings me to an article that now has a new heading, or the top article in the mail is lower on the list when I go to the site.

  70. Skarre-r and Bergensk r are the same thing, uvular r.

    Shows how crappy my ear can be, I guess. I listened to someone saying Bergen and I could have sworn he was using lingual [r], not any kind of uvular trill, fricative, or approximant. (I am no David M to distinguish by ear between laminal and apical consonants.)

    But then again, I suppose just because he’s saying Bergen doesn’t mean he’s from there.

    I’m not certain I really understand what rhotic is

    It means that historic r after a vowel is pronounced as a consonant. When I say radar, for example, I use exactly the same consonant at both ends of the word, whereas you use a consonant (the same consonant) only at the beginning.

    But the exact nature of this consonant makes no difference in calling an accent rhotic. Dutch and French radar likewise have the same consonant at both ends, German Radar only at the beginning, but they aren’t (in most people’s pronunciation) all the same /r/, and none of them are the English/American /r/.

  71. PlasticPaddy says

    The initial German r and Parisian r in Radar are the same in some German accent (I am not sure which one). Dutch were ruled by Spanish and pattern themselves in frugality after the Scots, so they roll their ‘r’s.

  72. John, thanks for such a succinct yet thorough description. Perfect! 🙂

  73. Trond Engen says


    Singer/songwriter Jan Eggum (b. 1951): En natt forbi
    Dance/soul singer Gabrielle (b. 1985): Fem fine frøkner
    Satirical TV show 5080 Nyhetskanalen

  74. Trond Engen says

    For those with little patience for entertainment, here’s a short clip as a dialect sample:


    (The speaker self-describes as somewhat at the posher end, but this doesn’t affect the r’s. It’s also where Jan Eggum’s lyrics in the song above.)

    For contrast, a sample of a speaker with the old western r:


  75. Thanks, Trond, those are very helpful!

  76. Trond Engen says

    For another contrast, the available sample most close to my own speech:


    (An eastern urban educated male speaker close to my age. Weird.)

    You’ll hear the apical r’s, retroflex l’s, and sandhi products thereof, that are absent from the western samples. You’ll also hear the difference between western high-pitch and eastern low-pitch stress.

  77. Lars Mathiesen says

    Groovy assimilation in varmt [ˈʋɑɳʈ] — Norwegian R’s are muchly strong, pulling labiodental to retroflex though they vanish in the process.

    (Maybe the dental stop helped before it was retroflexed itself, but still… the other way, /n/ > [m] before labial stop, is all over in Danish and I think the rest of Scandinavian).

  78. David Marjanović says

    Up to a point, Minister. Few Scots use that any more. For the most part they use the same /r/ as English and Americans, or at most the tapped /r/ of Spanish r rather than rr.

    My impression is that trilled /r/ remains very common in Scotland (where I’ve been twice, without interacting with a lot of people), while approximants (and losses) mostly show up when Scots go out of Scotland and address a broader audience.

    When it’s trilled, it’s probably apical* all the time, no matter how long it is. In Spanish and Basque the one-contact trill (“tap”) is apical like in Italian, Icelandic or Finnish, while the four-to-five-contact trill is laminal like in Slavic or Swiss German.

    * Except for that one MP who uses uvular trills almost, though not quite, all the time.

    Not that these sounds were unheard-of in England as little as a hundred years ago.

    John Bercow does this a lot to enhance his theatralic performance and so he doesn’t need to increase the volume of his voice quite as far as otherwise.

    indeed, that variety is more non-rhotic than the Queen, saying “Mistuh Adams” rather than “Miste-r-Adams”.

    White people in South Carolina traditionally go even further: as in Danish, /r/ only survives as a consonant if a stressed vowel follows, so it drops out of Carolina [kʰɛəˈlaˤːnə].

    A particular feature of rising and falling rhoticity is that the NURSE vowel is the first to go rhotic and the last to go non-rhotic.

    That may be because of the NURSE-CHOICE (nurse-noise) merger that the New York area used to have. Bernie Sanders’s stressed NURSE vowel, as in Bernie, is [ɹ̩], while his unstressed version, as in Sanders, is [ɐ]. I would not be surprised if he natively had the NURSE-CHOICE merger, learned which words were supposed to be NURSE, and put a mainstream American value for NURSE in them.

    I am no David M to distinguish by ear between laminal and apical consonants.

    I don’t claim to be good at this. Drop a few trills on me in the middle of a conversation, and I probably won’t be able to tell – not necessarily even whether they’re lingual or uvular.

    With other consonants it’s easier. I noticed something was off with my /t d n/ in English years before Justin Rye told me that the English ones are apical; then it suddenly made sense, and I pronounce them apical in English now.

    The initial German r and Parisian r in Radar are the same in some German accent (I am not sure which one).

    Almost all of them by now, as opposed to 70 years ago.

    …with the added complication that in France and most of Germany, /r/ is actually not a trill, but a fricative or approximant most of the time. This phenomenon has not reached Austria for example.

    Dutch were ruled by Spanish and pattern themselves in frugality after the Scots, so they roll their ‘r’s.

    Sociolingustically, uvular r (trill, approximant, fricative) has up to two features: it is urban middle class, and it is French. The first feature explains its wide and rapid spread. The second explains why it has not reached Flanders, where /r/ remains a laminal trill. In the Netherlands, where there’s no such need to emphasize not being French, uvular /r/ is probably universal now, except in the areas that use an English-style* apical approximant. (And some areas use uvular trills before vowels but apical approximants after vowels.)

    In Germany, the apical trill has largely become a marker of Bavarian ethnic identity; this may likewise be sociolinguistic resistance to a feature perceived as Prussian. In Austria, the uvular trill is very widespread now, though the laminal one can still be found, and I’m not aware of geographic or social connotations (even though Vienna went uvular only in the 1960s to 70s). Perhaps similarly, the areas with a laminal trill – Switzerland & surroundings – seem invulnerable to uvularization, except that Wikipedia says the Bernese upperclass traditionally trills uvularly.

    The uvular trill has even spread to the northern fringe of Slovenia.

    * Or Albanian-style, if you prefer, except that Albanian has a trill, too (spelled rr).

  79. @David Marjanović: Very few white people in South Carolina talk like that now, and from what the older locals have told me, that was never the majority white accent. The most famous example of somebody with that level of non-rhoticity was former governor and senator Fritz Hollings (although I cannot at the moment find an online recording of his famous “running out of courts” speech, where, as governor, he accepted the legal reality of desegregation and commited the state to following the rule of law). Supposedly, that accent was traditionally associated with the Carolina upcountry, although Hollings was a low country man (born in Charleston, died this past April in Isle of Palms).

  80. Trond Engen says

    Lars Mathiesen: Groovy assimilation

    Yes. This is a specifically Eastern/Central feature, shared with (some) colloquial Swedish. It’s generally seen as a consequence of the retroflex development of l and rd, but I’ve suspected influence on a more basic level from the low pitch in stressed syllables.

  81. John Cowan says

    Piotr told us five yearrrrrs ago that both he and Donald Tusk have idiolectal uvular /r/. Mark Shoulson can make all the (non-diacriticized) sounds in the IPA except [r]; I myself fail at [ɧ], at least on the assumption that it truly is a doubly articulated fricative.

  82. David Marjanović says

    [ɧ] is easy. “Just” give your tongue an S-shape.

    In reality, the pronunciations of the actual Swedish diaphoneme vary incredibly widely. I’ll look for a link tomorrow or so. Part of the reason may be that a lot of power is needed to articulate [ɧ].

    I can make [r], both kinds, but not at all reliably.

  83. Trond Engen says

    I’ll add that I wanted to provide a sample of Hat’s mother’s ancestral southwestern dialect too. Since the parish transformed into a hydroelectrically powered industrial town in the 20th century, and the dialect bevame a rregularized regional melange, I thought I’d use a neighbouring parish instead, but it’s nowhere on the ‘Net.

  84. My mom’s family was from Sauda, if anyone’s curious.

  85. David Marjanović says

    A non-rhotic trilling Scottish accent. Also, your is still in CURE, like for Schwarzenegger.

  86. John Cowan says

    All I hear there are taps. For me as well, your is CURE when contrastively stressed, NURSE otherwise, part of the general English system of (phonologically) strong and weak pronouns. Even [h] pronouncers will drop it in “I saw im/er”, for example, and “I want to go” for me is [əˈwɑnəgoʊ], with the I maximally reduced.

  87. Or, presumably, [əwɑnəˈgoʊ], depending on context.

  88. Yes.

  89. David Marjanović says

    All I hear there are taps.

    Single-contact trills, like most trills worldwide. Taps are supposed to be ultrashort plosives, and I don’t think such a thing even exists.

  90. David Marjanović says

    [ɧ] is easy. “Just” give your tongue an S-shape.

    In reality, the pronunciations of the actual Swedish diaphoneme vary incredibly widely. I’ll look for a link tomorrow or so.

    Start here and then dive down the rabbit hole head-first.

  91. “Plosives of unusual length? I don’t think they exist.”

  92. Lars Mathiesen says

    ɧ spelled stg. We got that deep earlier this fall already. (Not at David’s link, but in WP I think).

  93. Trond Engen says

    I should probably add something about ɧ. I think I’ve said before that it’s an all-mouth sound of breath, with friction taking place on any and all possible points of articulations at once, and that rounding, palatalization and other points of (co)articulation may gain precedence in anticipation of the following vowel. But I’m much less confident than I used to be. Linguists write very different things about it. One reason could be that the phoneme (or whatever) is being interpreted wildly different by speakers. Maybe it’s in the process of changing from phoneme to suprasegmental feature. The outcome could be +VOICELESS on the following vowel.

  94. Lars Mathiesen says

    Maybe /ɧ/ should be defined negatively, as an unvoiced fricative distinct from /f/, /s/, /ʃ/ and /rs/ = [ʂ] (except if you think [ʂ] sounds better). It more or less has to have some velar feature to achieve that, but the field is wide open.

  95. John Cowan says

    So it never merges with any of the other unvoiced fricatives, or do they move to get out of its way? This page from Possessive Suffix says the four basic ones are /f/, /s/, /ɕ/, /h/, which doesn’t match what you have.

  96. Lars Mathiesen says

    Well, for /ʃ/ read [ɕ], there is no /ʃ/ – /ɕ/ contrast so I picked the more traditional symbol. But that’s a nit. And yeah, /h/ is relevant, you need to keep your /ɧ/ more noisy.

    The R+alveolar > retroflex thing is a Svealand/Norrland feature, but that’s the variety I know best. In Stockholm you will indeed hear a few people realizing /rs/ and /ɧ/ the same, though it feels an affectation. (“_I_ don’t use that ugly sound!”). I do think that PS is right about [ʂ] being a preferred realization in general, but only where it won’t merge with the product of retroflexion.

    (Also I don’t know what happens with /r:s/ or with /r(:)ʃ/, /r(:)ɧ/ in Stockholm — I probably said them wrong if I tried. Herrskjorta ‘mens’ shirt’ might just have a [ʂ:] for ‘rrskj,’ but I’d probably go for something like [ɾʂ]).

  97. Stu Clayton says

    What is the Swedish word/expression for “hair shirt” ? German has härenes Hemd, which is not that close in sound to Herrenhemd.

  98. Tagelskjorta, apparently.

  99. Stu Clayton says

    GT, rendering part of that article into German for me, came up with the nonexistent word Trensenhemd. A Trense is a horse’s bit. For English it gave “bridle shirt”. Something to match a bridal gown ?

    Not sure what that’s all about. Dumb algorithms applied to dicey data.

    Most of the translation into German was correct. You’d think that syntax would cause more problems than nouns. Maybe it does in general. But perhaps you can think your way around a garble of words, each of which needs the others (syntax), whereas one wrong noun brings you to a screeching halt. Belief in indexical autonomy is hard to break.

  100. Lars Mathiesen says

    Sw tagel is hair from the mane or tail of a horse (also used for upholstery, designed to puncture the skin of shorts-wearing little boys when it inevitably escaped the covering) — but GT probably got its indexes crossed in some equestrian context. (Are there EU regulations on horsekeeping?). For repentance you don’t make your underthings from the hair of fluffy bunnies, you want something stiffer.

    Danish uses the compositional bodsskjorte, ‘shirt of boot’ (to use the cognate — betterment, not footwear).

  101. PlasticPaddy says

    Rather than betterment, do you mean penance (compare German Buße)?

  102. Lars Mathiesen says

    Yes, penance is the sense of bod here, but E boot has the older sense of betterment in to boot, for instance. (It does in fact have the same root as better). Danish has the collocation bod og bedring which is what you promise the judge — and a fine is bod and bøde.

  103. the pronunciations of the actual Swedish diaphoneme vary incredibly widely

    Nothing special in Hej gamle man as sung by ABBA (the word to listen for is kanske):


  104. For comparison, the same word as sung by Marie Fredriksson (around 1:22 in):


  105. A Russian take on it:

    Вы слышали, что есть такие необычные слова: «сыски», «хыхки» и «фыфки»?


  106. David Marjanović says

    For comparison, the same word as sung by Marie Fredriksson (around 1:22 in):

    That’s a very emphatic [xˁ].

    Where is it in the ABBA version? I don’t want to listen to the whole thing.

  107. Lars Mathiesen says

    0:49 and then I didn’t listen further. Sidetrack attempt: Note the obsolete masculine form gamle in the song title, I thought ABBA were too young for that. (Benny and Björn born 1945-46).

  108. David Marjanović says

    Huh, that’s just a [ʃ].

  109. Lars Mathiesen says

    That was actually what we were told to use for singing (in a choir).

  110. the obsolete masculine form

    I remember hearing Nobelpriset i litteratur för året 2005 tilldelas den engelske författaren Harold Pinter.

  111. för året

    I think it was på året.

  112. författaren
    Trying to understand Swedish from German, I first thought “what does ‘ancestor’ do here” (l parsed it as Vor-Vater). Then it hit me – this is Verfasser “author”.

  113. Variations between -e and -a

    In written and formal spoken language (and in the spoken language of
    South and West Sweden) –e is usual when describing a male person,
    although the –a form is becoming increasingly common.
    den unge killen the young boy
    Evas trevlige kusin Eva’s nice (male) cousin
    Note that –e often has a higher stylistic value than –a.
    hennes gamle make her elderly spouse
    cf. hennes gamla gubbjävel her old devil of a husband

  114. Andrej Bjelaković says

    Re: Scottish, the loss of non-prevocalic /r/s was noticed among young urban working-class speakers some 40 years ago.

    IIRC, first you have an alveolar tap/single contact trill with some pharyngealization, then it weakens to an alveolar approximant with pharyngealization, and finally you’re left with a pharyngealized vowel, so that e.g. ‘card’ is [kʰɑˤːd]

  115. PlasticPaddy says

    In Irish English (as in Irish) an epenthetic vowel is more likely to be added rather than dropping the sonorant. This is the case also for Scots Gaelic. See https://gaelicgrammar.org/~gaelic/mediawiki/index.php/Epenthesis
    Which urban centre do we mean here? Not Edinburgh,I expect☺

  116. Andrej Bjelaković says

    From Stuart Smith et al. 2007:

    “Loss of postvocalic /r/ was noted in working-class children in Edinburgh by
    Romaine (1978), with boys leading the change. Macafee (1983: 32) noted
    similar changes in ‘adult speakers in Glasgow’.”

    She has done a lot of work on the subject in the last 10-15 years. See here for example:

  117. PlasticPaddy says

    See https://www.abdn.ac.uk › PFRLSU
    Warren Maguire has a paper about epenthesis based on analysing fieldwork notebooks. It appears to find epenthesis in both cities, looking at the maps. Since, as he says, fieldwork informants are usually old working-class people, maybe the words are different or the dropped r is seen primarily among younger speakers. Or maybe I am reading the maps wrong ☺

  118. Andrej Bjelaković says

    Indeed, as far as I can see, he uses old Linguistic Survey of Scotland recordings, made in the late 1950s?

    Also, briefly looking at the paper you mention, he talks about how often epenthesis IS there, not how often /r/ ISN’T there. But again, if these recordings were made back then, and involved mainly older people, then I’m sure no non-rhoticity was present to begin with.

  119. PlasticPaddy says

    Thanks. I will listen out for this ????

  120. Then it hit me – this is Verfasser “author”.

    ~~ flashbulb ~~

    And what are the writings of a Verfasser, etymologically? Far-fetched!

  121. David Marjanović says

    “Far” is fern, though (one of the many, many, many words where -er(-) became -ar(-) in English).

    Wiktionary does tentatively confirm that fetch is a root cognate of fassen (apparently it’s the *j-present or something).

  122. Yes, “forfetched” might have been better, but I like it this way.

Speak Your Mind