I imagine most of you are familiar with the old wheeze about fish being spelled ghoti, with gh pronounced as in laugh, o as in women, and ti as in nation. It’s regularly attributed to Shaw, but no one has ever found it in his writings, and it turns out, as reported in an invigorating Language Log post by Ben Zimmer (now Executive Producer of—congratulations!), that that’s because it goes back before he was born, being attested in a letter dated December 11, 1855, to Leigh Hunt from his publisher Charles Ollier:

And here an experiment in orthography, which it may amuse some of our readers to carry further at this season of puzzles and charades, and kindred jovial perplexities:—”My son William has hit upon a new method of spelling Fish. As thus:—G.h.o.t.i., Ghoti, fish. Nonsense! say you. By no means, say I. It is perfectly vindicable orthography. You give it up? Well then, here is the proof. Gh is f, as in tough, rough, enough; o is i as in women; and ti is sh, as in mention, attention, &c. So that ghoti is fish.”

As Ben says, “it actually makes sense that ghoti made its earliest appearance in the mid-nineteenth century, when English orthographic reform was gaining popularity”; he quotes some far more ponderous examples of the same sort of jovial respellings from the period (showing, incidentally, that as of 1845 postvocalic /r/ was still pronounced in educated usage).
One thing that particularly pleased me was the discovery that the erroneous attribution to Shaw comes from Mario Pei, who Ben calls (with perhaps excessive kindness) “not always the most reliable source when it comes to language-related information.” Pei, like Bryson today, is enjoyable to read but not to be taken seriously as a source of facts.
Oh, and the Log now has comments (again)! Well done, chaps.


  1. rootlesscosmo says

    Along the same lines is “psoloquoise” as a way to spell “circus”–“olo” as in “colonel” and “oise” as in “porpoise,” and the initial “p” is silent, as in “pshrimp.” (Hat tip: P. G. Wodehouse.)

  2. Who preonoun ces ‘women’ as ‘wimmen’?
    I pronounce the ‘o’ in ‘women’ the way I pronounce ‘oo’ in ‘good’. Somebody saying ‘wimmen’ would sound very uneducated to me.

  3. ‘preonoun ces’ meaning ‘pronounces’, of course.
    (As keyboards get smaller and smaller my typing gets weirder and weirder).

  4. I don’t know accurate the accents really are in the recent HBO “John Adams” series, but the actor playing King George III did pronounce post vocalic /r/ (and pronounced “I” almost like modern “oy”), so the producers were clearly trying to get across to viewers the idea that the accents of those days were not what we hear today.
    Also I find it personally interesting that young William equated the vowel of “fish” with the first vowel in “women.” Personally I think I produce “fish” with a more fronted sound, and “women” as almost a schwa, but maybe I’m lazy.

  5. Bob Helling says

    Stephen, I think just about everyone pronounces women (the plural of woman) as wimmen. It sounds like you are pronouncing woman.
    I thought it was interesting that in the quoted text they talk about “sh, as in mention, attention.” I pronounce the ti in those words as ch.

  6. Stephen,
    How do you distinguish between the singular “woman” and the plural “women”?

  7. Bob Helling says

    Okay, I already regret the “just about everyone comment” as vanya shows there can be variations. But I can’t remember hearing it ever pronounced other than wimmen.

  8. “showing, incidentally, that as of 1845 postvocalic /r/ was still pronounced in educated usage”
    Note that the same word also supports a different pronunciation of the long vowel sound than RP (“nEEther” rather than “nEYEthuh”)

  9. Like Bob, I can’t remember ever hearing it ever pronounced other than “wimmin” (I’d write it with two i‘s, not an i and an e). The OED has /wʊmən/ for “woman” and /wɪmɪn/ for “women,” and this matches how I’ve always heard them said.

  10. Bob: the American Dialect Society mailing list thread on Matthew Gordon’s discovery has some further discussion on the pronunciation of “ti” in “mention”/”attention”.

  11. Who preonounces ‘women’ as ‘wimmen’?
    As Bob says, just about everyone.
    I pronounce the ‘o’ in ‘women’ the way I pronounce ‘oo’ in ‘good’.
    Are you a native speaker? I have never heard, or heard of, anyone pronouncing it that way. The /i/ comes straight from Old English wifmen; to quote the OED, “From at least the 16th century, the only variety in the pronunciation of the pl. has been in respect of the quantity of the first vowel, which was either short or long in the 16th and 17th centuries.” The word is spelled with -o- by analogy with the singular.

  12. A few years ago, after first coming across ghoti, for a period of a few months, when I would think of it, I was filled with a strong feeling of dread, as if the world was somehow at a tilt, as if the angles of a Euclidian triangle no longer equaled 180 degrees.
    That passed.
    Since then my theory has been that the originator of ghoti was none other than the Great Cthulhu.
    Go ahead, prove me wrong.

  13. Jeremy Gaunt says

    Reminds me of an old joke about a rather backward U.S. student who was told he could have a university football scholarship if he could spell at least one letter right in the word coffee. He came up with KAWPHY

  14. mollymooly says

    Concur with the mob on wimmin.
    For me the first vowel in woman is that of STRUT rather than FOOT: the word rhymes with comin’ (as in, round the mountain). I was amazed when I first read in the OED that this is “vulgar dialectal”. Perhaps it is standard in Ireland, though I am not conscious of English or Americans using a different vowel. To my ears, /wʊmən/ sounds Scottish (or perhaps I mean, a noticeably Scottish woman — Kirsty Wark, say — sounds like /wʊmən/). I guess the preceding /w/ is messing with the phones, or with my ears.

  15. John Emerson says

    And tonderceo would be pronounced “to” (from Cholmondeley and Worcester).

  16. Cholmondeley and Featherstonehaugh
    = chumly & fanshaw
    Makes one wonder if English is English, or if the world hasn’t been taken for a ride by the likes of Prunk and Strine.
    Makes one wonder, also, if those historical linguists who relate phonology directly to genetical language trees really know what they’re going on about.

  17. But:
    tonderceo would be pronounced “to” (from Cholmondeley and Worcester).

  18. rootlesscosmo says

    tonderceo would be pronounced “to” (from Cholmondeley and Worcester).

    “onde” in Cholmondeley is silent; so is “rce” in “Worcester.” Thus t[silent letter string][silent leter string]o” = “to.”

  19. No, no. As a Mass resident I assure you, the “ce” in “Worcester” is pronounced. It’s the “s” that is silent.

  20. OT:
    I listened to an interview you did for NPR where you talked about your book about “bad words” and insults in different languages. What is the name of the book?? I can’t find the post and I can’t recall the book’s title.

  21. I grew up near a Worcester in South Africa. In Afrikaans it’s pronounced like in English (silent rce), except with a fricative w. This also happens to be a homonym of the Afrikaans for “rougher”, which, in this Worcester, is very apt.

  22. Daniela: here‘s the LH post, and here‘s the Elwin Street page. If and when the US edition appears, I’ll put a link in the LH sidebar.

  23. This struck me as very funny back when I was learning how to spell (or maybe immediately after I had learned how to spell): The spellings of “woman” and “women” differ in only one letter, the vowel of the first syllable; but the pronunciations of the two words (in my idiolect) differ only in the vowel of the second syllable. I say “wommin” and “wimmin” — as near as I can hear, I pronounce the second syllables of the two words exactly the same.

  24. David Marjanović says

    Since then my theory has been that the originator of ghoti was none other than the Great Cthulhu.

    Iä! Iä!
    However, it goes without saying that ghotI’ is “the most general word for a fish-like creature” in Klingon. Pronounced [ˈʁotʰɪʔ], except that the sources I’ve read and heard contradict each other on whether H and gh are velar or uvular.

  25. Cthulhu Fthagn!

  26. David Marjanović says


  27. Cthulhu will eat the prescriptivists first.

  28. The spelling and pronunciation of ‘woman/women’ (spelling and pronunciation distinctions reside in different syllables) reminds me of the Japanese borrowing of the English words ‘trouble’ and ‘travel’.
    Trouble = トラブル (toraburu)
    Travel = トラベル (toraberu)
    As you can see, the distinction in the English is carried by ‘oub’ vs ‘av’ (i.e., ‘ub’ vs ‘av’). The distinction in Japanese mirrors the spelling distinction in the last syllable of the English (‘le’ vs ‘el’). It’s nothing short of amazing how katakana transliteration conventions completely miss the real sound distinction and make up for it with a non-existent one!

  29. David Marjanović says

    Cthulhu will eat the prescriptivists first.

    This is a good reason to become a prescriptivist!!!

  30. Thippu Sulthan says

    on seeing this page every one to discuss woman women pronunciation the word women is “v may in” woman”voomen”

  31. Marc Okrand pronounces Klingon gh as a uvular fricative, and so does Mark Shoulson (aka Seqram), but there’s nothing wrong with Okrand’s ears: in the Klingon Dictionary he says it is a voiced velar fricative, and a voiced velar fricative it must and shall be.

  32. Presumably the dialect learned by Okrand and Shoulson differed in this respect from the standard described in the Dictionary.

  33. Naah, I’ve talked to Mark about it, and he just doesn’t do velar fricatives; his Klingon has a Hebrew accent. It’s still better than Michael Dorn’s.

  34. I’ve seen a lot of informal texts use “velar” when they mean “uvular” (the same with “alveolar” and “dental”), so I would have expected [χ] anyway from that description (seeing as how [χ] is perceived as a “harsher” sound, more befitting of Klingon).

  35. January First-of-May says

    The Ellis respellings make me think of (Scottish and) Irish Gaelic – though I’m not sure if they’re in any way deliberately influenced by Gaelic orthography or just accidentally similar.

    Essentially obligatory, and somehow not linked before: George Bernard Psschaughal. I particularly like the punchline.

  36. David Marjanović says

    What Lazar said. I’ve even seen a description of the Klingon “velars” in some authoritative-looking source as “velar, like in Yiddish”, where they’re uvular.

  37. Velar and uvular were often confused under the old-fashioned word “guttural”.

  38. Lars (the original one) says

    the punchline: Given the choice between [o] and [ɑ] for English [ɔ], I’m pretty sure the Swedish Academy of the Sciences would spell it Tja. Tjå would probably work in Norwegian, and Sjår in Danish. (But none of these language has a [ʃ] so [ɕ] has to suffice).

  39. The ipsissima verba:

    gh This is not like anything in English. It can be produced by putting the tongue in the same position it would be in to say English g as in gobble, but relaxing the tongue somewhat and humming. It is the same as Klingon H, but with the vocal cords vibrating at the same time.

    Now since no one has ever claimed that English /g/ is uvular in gobble, I submit that that describes a voiced velar fricative. Technically it could be an approximant, but nobody says that, and the word “somewhat” speaks against it.

    It’s true that Okrand says H is like the ch in L’chaim!, so there is a bit of contradiction here. But he also says it’s like the ch in Bach. The back vowel in that word may pull the fricative toward a uvular pronunciation, but I think Okrand was choosing the word for its familiarity (to anglophones), not its uvularity. His other examples of H are Tijuana and Baja California, which are clearly velar fricatives in Mexican Spanish when they are not plain [h].

    In short, the uvular sounds of Klingon are the voiceless stop q and the voiceless affricate Q [qX], not the fricatives.

  40. David Marjanović says

    But he also says it’s like the ch in Bach. The back vowel in that word may pull the fricative toward a uvular pronunciation

    Back or just open, it does pull it all the way to a uvular pronunciation in most of Germany. [x] is actually a rare sound in these accents, more or less limited to the position after [uː].

  41. Stephen C. Carlson says

    the punchline: Given the choice between [o] and [ɑ] for English [ɔ], I’m pretty sure the Swedish Academy of the Sciences would spell it Tja.
    That sounds about right, and even better for Americans with the cot-caught merger. The punchline has “Skjå”, which would be /ɧo/, and which my two Swedish teachers pronounced differently (one was from the north, the other from central Sweden).

  42. In Russian it’s Шоу (= “show”).

  43. David Marjanović says

    More on Bach: WP-de on Ach-Laut says “velar or uvular”; its talk page cites the IPA Handbook as saying it’s [x] after /oː ʊ uː/, [χ] after /a aː ɔ/. WP-en onich-Laut and ach-Laut appears to cite the same source: “According to Kohler,[77] the German ach-Laut is further differentiated into two allophones, [x] and [χ]: [x] occurs after /uː, oː/ (for instance in Buch [buːx] ‘book’) and [χ] after /a, aː/ (for instance in Bach [baχ] ‘brook’), while either [x] or [χ] may occur after /ʊ, ɔ, aʊ̯/, with [χ] predominating.” The more recent one of the two references in footnote 77 is this paper, which describes the pronunciation “of many educated Germans in the North” and ascribes [χ] to Dach on the first and only accessible page. (Unrelatedly, there are two embarrassing cases of β typeset for ß there.)

    As you can guess, none of this is universal. You can hear Bach with [x], you just need to know where to go. I lack [χ] altogether, and because my /a/ is a front [a], the /x/ in Bach is even slightly fronter than the one in Buch.

  44. January First-of-May says

    In Russian it’s Шоу (= “show”).

    As in the classic Russian “blonde girl” joke:

    – Вам нравится Бернард Шоу? (Do you like Bernard Shaw?)
    – Не знаю… а по какому каналу оно идёт? (Don’t know… what channel is it on?)

    (…For some reason this comment didn’t originally seem to go through, and my later attempts to post it got hit by a system message telling me that I said that already.)

  45. Tjå would probably work in Norwegian

    No. Sjå is how we hear it. Skjå is homophonous, except maybe for some conservative western dialects, but that’s an unnecessary complicatification. The tj-sound merged with the kj-sound a few generations ago, and together they are merging into the sj-sound as we speak, but the latter merger is being frowned upon and may well end up being reverted.

  46. But none of these language has a [ʃ] so [ɕ] has to suffice.

    To a normal anglophone there is no perceptible difference between the two even at the allophonic level.

  47. Nor to me, unless I consciously force a difference in point of articulation. They are both well within the range of a Norwegian sj-sound.

    (I used “sj-sound” etc. as in Norwegian folk phonology to avoid a discussion of phonemenology. Eng. Wikipedia on Norw. phonology uses ʂ and ç. Nynorsk Wikipedia list three phonemes, ʂ, ʃ, and ç.)

  48. Lars (the original one) says

    which my two Swedish teachers pronounced differently — they would. The ‘sje’ sound is very variable even in Stockholm, ranging from [x] to [ç] or even [ʂ] between speakers, words and registers, just as long as it has sufficient contrast to [ɕ]/[ʃ] (the ‘tje’ sound which is always -dental/+sibilant/-retroflex).

    There may even be people who can do a ‘real’ [ɧ] which according to the IPA is simultaneous [ʃ] and [x], but I don’t know how to hear if they do, and I certainly got away without trying.

    But that’s central Sweden, many other areas have complete merger between the ‘sje’ and ‘tje’ phonemes as [ç] or [ʃ], the North and Finland among them. Or so I’ve been told.

  49. Stephen C. Carlson says

    But that’s central Sweden, many other areas have complete merger between the ‘sje’ and ‘tje’ phonemes as [ç] or [ʃ], the North and Finland among them. Or so I’ve been told.
    My Swedish teacher from the north had a dictation with the words kärna and stjärna. She swore they sounded differently, but I could never tell them apart. I even tried lip reading and was still hardly better than 50/50. The other teacher, yes, I had no trouble distinguishing her [ɕ] and [ɧ].

    Around Uppsala at least, the [ɧ] sound had a lot of labialization, almost like /xʷ/ or even /hʷ/. (Unfortunately, my ear is not trained particularly well for phonetics, so this is just impressionistic). For some people, the sound had different allophones depending on where in the word it occurred.

    I’ve read that outside of Sweden, the sj-sound [ɧ] is usually taught as /ʃ/ in Swedish classes, but you’d hardly hear that in Uppsala (aside from allophones), unless the person is from elsewhere, for example, the prime minister Stefan Löfven, who was raised near Sollefteå.

  50. David Marjanović says




    Ooh, that’s even better.

  51. David Marjanović says

    MOAR! u!vu!lar! fri!ca!tives! in! Kling!on!

  52. David Marjanović says

    Maps of the phonetic details of the ach-Laut: lots of uvularity even behind /uː/.

    (Warning: Bruch doesn’t have /ʊ/ for everyone; it has /uː/ for me… I’m not sure if I have the sequence /ʊx/ anywhere, in fact, other than Kurt Kuch whose name I only know from reading.)

Speak Your Mind