I haven’t had a lot of pho, the Vietnamese noodle-and-meat soup, but what I’ve had I’ve liked. The name is a notorious problem (for English-speakers): it looks like it should be pronounced like the word “foe,” and lots of people say it that way, but the Vietnamese word phở actually has a mid-vowel that sounds like the vowel of “fun” (you can hear it said at that Wikipedia article by clicking the “listen” symbol), and those English-speakers who know that can sound a little silly trying to reproduce it in English (and, worse, can sound supercilious if they feel impelled to “correct” those who say it like “foe”). But never mind that; where is the word from, what is its etymology? The OED (entry created 2006) says “< Vietnamese phở, perhaps < French feu (in pot au feu),” and AHD says the same, but it seems odd that such a basic Vietnamese dish should have a name of foreign origin. Not absurd, mind you, or even unlikely (cf. “hamburger”), but odd. So I was interested to see an entirely different origin proposed in Andrea Nguyen’s The History of Pho; it’s well worth reading if you have any interest in the soup itself, but here I’m focusing on etymology. She says it began (around the turn of the 20th century) as a beef noodle soup called nguu nhuc phan:

So how did nguu nhuc phan become pho? It is likely that as the dish caught on, the street hawkers became more competitive and abbreviated their distinctive calls as a means to attract customers. “Nguu nhuc phan day” (“beef and rice noodles here”) was shortened to “nguu phan a,” then “phan a,” or “phon o,” and finally settled into one word, pho. In a Vietnamese dictionary published around 1930, the entry for pho defined it as a dish of thinly sliced noodles and beef, its name having been derived from phan, the Cantonese word for flat rice noodle. It’s been suggested that pho arose because when phan is mispronounced or misheard, it can mean “excrement.”

The term pho is not French in origin, despite claims that the pronunciation bears resemblance to feu (fire in French, as in pot au feu).

Intriguing, but Nguyen is a food writer, not an etymologist, and I don’t know enough about Vietnamese to have an informed opinion. As always, I welcome all thoughts from the Varied Reader.


  1. I have nothing to contribute on the Vietnamese side of the story, but the Cantonese word that Nguyen writes as phan would usually be transcribed as fan2 (pronounced fěn in Mandarin; not to be confused with Cant. faan6, Mand. fàn, meaning “cooked rice”).

  2. J.W. Brewer says

    I think I’ve actually eaten at a place called something like “Pot Au Pho,” although I expect the immigrant ESL-speaking proprietors may have gotten naming advice from someone else with better cultural insight as to what wordplay would seem clever/memorable to a largely white college-town customer base.

    Re “not absurd,, or even unlikely,” certainly no more unlikely than that a major Japanese dish should have a name of ultimate Portuguese origin, i.e. tempura.

    Ms. Nguyen may well be right, but her account would be more convincing if she didn’t concede that neither the dish nor the name predated the period of French colonial rule.

  3. I can go on about pho for a long time 🙂 but just wanted to add a tidbit about pronunciation hints: there is a Las Vegas joint called Pho King to help us with that.

  4. J.W. Brewer says

    My now-wife (who ate their more frequently) tells me that it was indeed Pot-au-Pho, but it closed down maybe 3 or 4 years ago, so the cleverness of name was not sufficient to overcome the vicissitudes and tight operating margins of the ethnic-restaurant business.

  5. In Vietnamese markets the dry noodles to cook at home are sold as pho. I think Nguyen’s etymology makes sense.

  6. We may say in general that most restaurants, at least in the U.S., have already sunk under a load of irrecoverable debt before they have even opened. In my view, the only way for a restaurant to remain open in the long term is to become so famous that they can survive indefinitely on pure tourist curiosity value, in which case the food served can be literally white worms in crankcase oil for all anyone cares.

  7. I want to be the first to mention pho-nology, and so, this paper, by Kang, Phạm and Storme, discusses French loanwords in Vietnamese. French /ə/,/ø/ and /œ/ indeed all correspond to Vietnamese ơ [ɤ] in loanwords, e.g. docteur > đốc tơ, so at least that part is consistent with the feu etymology, even if it’s ultimately false.

  8. We may say in general that most restaurants, at least in the U.S., have already sunk under a load of irrecoverable debt before they have even opened. In my view, the only way for a restaurant to remain open in the long term is to become so famous that they can survive indefinitely on pure tourist curiosity value, in which case the food served can be literally white worms in crankcase oil for all anyone cares.

    I presume you’re talking about fancy restaurants; profit margins are always slim, of course, but plenty of unpretentious neighborhood places survive for decades. Two of my favorites from my life in NYC: Luigi’s on 8th Ave. (try the baked ziti marinara, it’s not on the menu but it’s garlicky and delicious!) and Balkh Shish Kabab House in Astoria (my wife and I used to order out from there all the time, and still talk longingly about their green sauce). Luigi’s had been there long before I arrived in 1981 and I hope it will be there after I am laid to rest.

  9. The etymology of phở is actually simple: it is a loanword from 粉 (Middle Chinese pjunX, Mandarin fen3; means “powder” or “flour”, but also “noodle”) from a variety from Guangxi which lost final consonants (I cannot be more specific about which variety precisely, however without some digging).

  10. If it’s that simple, why are they dragging in French and unlikely-sounding phrase-shortenings?

  11. I presume you’re talking about fancy restaurants

    One of my acquaintances is a “serial restaurateur” who keeps starting unpretentious Caribbean restaurants (which then fail). He’s on 4th iteration now. With number 2, he lost his home, lived from couch to couch of friends. Number 3 failed more or less painlessly, in comparison. Now he’s joining forces with a more successful eatery which is nevertheless on hard times, and the prospects already look dim. It’s a calling, I suppose, to open restaurants which just look conceptually so wrong vis-a-vis public needs, yet irresistible for the owner’s inner system of values.

    If I were dreaming of a pho eatery, I’d say, if the Saigonese could add all their veggies, earn the Northerners’ contempt for that, yet succeed … then I should go a step further. Thinly sliced red and yellow peppers and mustard greens, anyone? Facebook-grade soup should be colorful, right? But then it’d fail 😉

  12. @languagehat: because nearly all Chinese loanwords in Vietnamese belong either to the Sino-Viêtnamese layer, or to even earlier borrowings; it is not usual to have a non-literary phonetic borrowing from a non-standard variety of Chinese, and this reading does not superficially resemble the readings in the other layers: 粉 is phấn in Sino-Vietnamese, bún in an earlier layer.

  13. Very interesting, thanks!

  14. The pronunciation is indeed tricky, or risky, as even the youngest English speaker knows. My seven-year-old says, “Pho’s not a bad word, right, Papa?”

  15. I’ve had pho at a pho place, but I still haven’t heard the word uttered too much – so I wonder to what extent Americans (in my neck of the woods, or in general) prefer “fuh” or “foe”. Maybe it’s a gyro situation.

  16. Maybe it’s a gyro situation.

    My thought exactly. (“Gyro” at LH.)

  17. David Marjanović says

    I know an American who makes an effort to pronounce it as [fə]; I don’t mean the STRUT vowel, I mean [ə]. Because that’s against English phonotactics, the word often ends up somewhat… aspirated maybe.

  18. SFReader says

    There is a Mongolian dish called пүнтүүз ‘puntuuz’ (from Chinese 粉条子 fěntiáozi, potato noodles).

    I searched on the web and discovered that Mongolians use “фо шөл” for Pho soup. They should call it ‘пүн шөл’, instead.

  19. The phonology of English prohibits open syllables with /ʌ/. They must be checked (ending in a consonant). That’s probably why ‘pho’ /fʌ/ seems off to English speakers. The interjection “duh” seems to be an exception.

    David Marjanović wrote: “I know an American who makes an effort to pronounce it as [fə]; I don’t mean the STRUT vowel, I mean [ə]. Because that’s against English phonotactics, the word often ends up somewhat… aspirated maybe.”

    An aspirated /f/? Is that just an idiosyncratic affectation, an attempt at sounding foreign? Americans don’t have [ə]; we pronounce the STRUT vowel and SCHWA with the same vowel quality.

  20. marie-lucie says

    Mark: Americans don’t have [ə]; we pronounce the STRUT vowel and SCHWA with the same vowel quality.

    This may be true when stressing SCHWA, especially by itself (eg to demonstrate the sound) but not for the unstressed vowel. At least that is my impression from paying attention to native speakers.

  21. Eli Nelson says

    For me, I’m pretty sure all putative word-final schwas, stressed or unstressed, are best transcribed with the “cut” vowel /ʌ/ (or phonetically, [ɐ]. There is a pretty clear difference in quality between my pronunciation of “Rosas” and “Roses.” While it’s true the latter is close to /ɪ/ (and pretty much indistinguishable from it), I’d prefer to transcribe it as /ə/ because it is also quite close to the vowel in “versus.” I’m not totally sure if I make the “rabbit-abbot” distinction; in any case, even if these vowels are not neutralized, it’s close enough that I cannot reliably hear the difference. So I’d transcribe the final vowels of both “rabbit” and “abbot” with /ə/, but for “Rosa” I’d use /ʌ/, and I’d also use this for “duh” and “pho.” Although this vowel may be raised a bit when it’s word-final and unstressed (since it does not contrast with schwa in that position), I think it is generally lower and longer than a true schwa (like in the first syllable of “police”).

  22. For me, [ʌ] and [ə] are in complementary distribution, and so only one phoneme (which is another way of saying that [ʌ] always undergoes vowel reduction when unstressed), but they are distinctly different in quality.

    I finally figured out whether I have the abbot/rabbit merger (it’s hard to hear, because unstressed vowels are out of focus) by listening to myself say chicken, where the unstressed vowel is traditionally [ɪ] despite the spelling. I definitely have the merger; if I didn’t, the two vowels would have the same quality.

  23. Yeah, there’s definitely more than one quality in play, with /ə/ being realized as [ə] non-finally but [ʌ] finally. (For me, at least, I think final means final: as far as I’m aware, my Rosa’s and roses are homophonous.) And I suspect this might be the reason why I’ve seen so many self-transcriptions by Americans, whether in IPA or fauxnetics, that indiscriminately use /ɪ/ for non-final schwas.

    @JC: Indeed, I don’t think you could round up many people on this continent for whom chicken doesn’t rhyme with thicken.

  24. As I pointed out before, the Chick-Fil-A cows who hold up signs reading “EAT MOR CHIKIN” indicates that whoever wrote that (presumably not a cow) did not have the merger.

  25. No, I doubt that. Like I said above, there’s a notable tendency for Americans to identify medial schwas with /ɪ/: I’ve seen numerous cases where people will self-report their pronunciations as, for example, “abbit” and “rabbit”, when both would likely equate to RP abbot.

  26. I’ve heard it said that “Bánh mì” comes from “pain de mie”, but the Wikipedia entry on Bánh mì says not. But I think Bánh does comes from pain.

  27. Isn’t 餅 more likely than pain? bánh covers a lot more semantic ground than just bread.

  28. David Marjanović says

    Why would a French [p] be borrowed as an implosive b [ɓ ~ ʔb] into Vietnamese, as opposed to an unremarkable p [b̥]?

    An aspirated /f/?

    No; an attempt to make the vowel checked by keeping it short, which appears to be accomplished by saying the whole syllable faster.

    Is that just an idiosyncratic affectation, an attempt at sounding foreign? Americans don’t have [ə]; we pronounce the STRUT vowel and SCHWA with the same vowel quality.

    For many of you it’s the same phoneme, but with clearly distinct allophones (stressed vs. unstressed, or stressed or final vs. unstressed non-final). This is why thummers exist.

    I should have mentioned that the person in question has some phonetic training; while [ɤ] may be beyond them, stressed [ə] as a one-word phoneme evidently is not, or not quite.

  29. There is, as far as we know, only one thummer, and his home language may not even have been English. However, there is a divergence between those who say stressed the with the FLEECE vowel before a vowel and the STRUT vowel before a consonant (effectively a closed syllable), and those who use the STRUT vowel in all cases.

    For me, unstressed them is replaced by hem /əm/, dating straight back to Middle English. (No, I don’t believe in magic /ð/-elision in just one word.)

  30. David Marjanović says

    I think I’ve read of more thummers, but Google doesn’t find them.

    Thanks for reminding me of The Simpsons, STRUT-KIT-DREEEEESS.

  31. Sino-Vietnamese for 饼 is bính, but bánh is probably from another layer of borrowing, though I will need some research to be sure.
    @David: there is no p- in Vietnamese, proto-Viet-Muong *p and *b both merged as ɓ, with high tone / low tone (the same with *t and *d, then a new t was created from *s and other onsets).

  32. David Marjanović says

    Oh! Oops.

  33. So, I’m just a lonely Western Vietnamese-learner, but I enjoy reading the rare linguistic articles on tiếng Việt. The pot-au-feu etymology has always struck me as unsatisfying because the Vietnamese also have a (vastly superior) equivalent of that dish, called bò kho. There is no reason that I can see for a completely non-French dish to borrow a word from French when it seems unrelated gastronomically.

    Now, there was something else mentioned upthread which was talking about how phở noodles are sold as such in Asian markets. This isn’t surprising; there are several different types of noodles used in Vietnamese cooking and the name of dishes usually include the name of that noodle (or perhaps, in some cases, the noodles take on the name of their signature dish). For example, bún usually refers to rice “vermicelli,” and is used in bún riêu, bún chả, bún măng vịt, and other dishes. (There is some variety, or inconsistency, as bún bò Huế noodles are thicker than other bún.)

    As for the pronunciation, it helps to remember that “official” and “actual” vary by region. My family (native speakers from the south) say it as a sort of breathy, questioning sound: fuh?. Native speakers from the north seem to say it with a little bit of vocal fry and a more “curled back” sound. It’s hard to describe.

    I will confess that people pronouncing it foe grates on my nerves when I hear it.

  34. Thanks for that well-informed comment!

  35. David Marjanović says

    Recently I found a place that rendered the ở as a cup with steam rising from it. 🙂

  36. Bánh mì is from 餅 (bánh is just an alternative reading of bính just as 正 can be chính or chánh) and 麵 (some Chinese branches are rid of the nasal ending giving rise to mì).

    As for phở, it is more plausibly from 牛肉粉 with 粉 losing the final nasal just as 麵 and 餅 have Vietnamised forms being mì and pía respectively. The durian bean cake famously derived from in and around Sóc Trăng is called bánh pía (also called bía), incidentally both syllables being derived from 餅. Quickly going to back to phở, one key thing to note is that while French loan words of -e/eu/œu in Vietnamese often become -ơ (chemise > sơ mi, beurre > bơ, sœur > sơ), there isn’t a single known loan word from French that has the hỏi or ngã tones. Furthermore if it were from pot au feu then you’d expect it to keep all 3 syllables as something like pót/pọt/pốt/pột ô phơ or pó/po/pô tô phơ.

    As for how to pronounce phở, the Northern tone dips and rises slightly while the Southern tone dips but rises sharply. Southerners also offen add a glide to the end.

  37. Very interesting; thanks for the information!

  38. David Marjanović says

    Ditto. I had the tones completely confused somehow.

  39. Although no one is sure exactly how pho came about, most believe that villagers saw an opportunity in in 1898 when French colonialists started constructing the Nam Dinh textile plant. French technicians and thousands of workers flooded into the region to work on Indochina’s largest silk plant, and these two soup dishes were likely combined and modified to cater to French tastes.

    “The French like beef so the villagers tweaked it by adding beef to the bánh đa cua,” Vuong said. “Also, they were given the discarded bones by the French who came to build the textile plant, so they took them back and boiled those to make the broth. They told us that was how pho was created.”

    As construction workers moved from Nam Dinh to Hanoi to work on the Long Bien bridge project, the nourishing noodle soup quickly spread beyond the village. Carrying portable pho stalls on poles over their shoulders, impoverished villagers followed the construction workers, making a good income from selling pho, and the meal-in-a-bowl soon became a favourite for people living in the capital.

  40. “feu” – this interested me because my sister lives with her son and Vietnamese daughter-in-law who is unusually forthright and candid about correcting mispronunciations by her mother-in-law, to the point where my sister has grumbled in semi-jest “where’s that Asian deference to the aged?” After having said d-i-l resident and correcting her for 6 months (and cooking pho regularly), my sister now pronounces it pretty much exactly like “feu”

  41. how to pronounce phở


  42. Since this post, AHD has changed their etymology for pho to the Cantonese source described above, including the ancestry of the noodle word:

    [Vietnamese phở, from alteration (possibly to avoid similarity to phẩn, excrement, or by protraction of the vowel in the cries of street vendors) of phấn in nhục phấn, meat and noodle soup (loan translation of Cantonese yuk6 fan2 : yuk6 meat + fan2 flour, rice noodle) : nhục, meat + phấn, flour (from Middle Chinese pun', from Old Chinese *mə.pənʔ, husked rice; perhaps akin to Tibetan dbur, to smooth).]

    That change was already made on the earliest Internet Archive capture of the page, dated 17 July 2018.

    Guillaume said: “粉 is phấn in Sino-Vietnamese”

    According to Wiktionary, phấn also exists in Vietnamese as a separate word, apparently from the “powder” sense of the Chinese word: “1. (cosmetics) face powder; 2. (piece of) chalk; 3. (botany) pollen”.

  43. That’s great, thanks for finding and sharing it!

  44. Perhaps too obvious to mention: this Cantonese word is also the “fun” in beef chow fun.

    Wiktionary also tells me that the same character is used in Chinese to transcribe the English loanword “fan”, as in fan of a person, and Language Log confirms: Chuānfěn 川粉 = fan of Donald Trump.

  45. David Marjanović says


    That seems to have â instead of ơ…?

Speak Your Mind