The perils of pronunciation

« previous post | next post »

Distinguishing between "four" and "ten" in rapid, slurred Modern Standard Mandarin (MSM) is not always easy:  sì 四 vs. shí 十.  Try saying sìshísì 四十四 ("forty-four") quickly and it starts to feel like the beginning of a tongue twister.  Now, when speakers from the various topolects, even within the so-called Mandarin group, come together and tones, vowels, and consonants start flying off in all directions, things can become still hairier and sometimes even costly.

Much to his chagrin, a supermarket manager from the southeastern province of Guangdong working in the southwestern metropolis of Chongqing found this out the hard way.

… Tan Jiayue, the owner of a supermarket in the huge southwest Chinese metropolis, called a printing house to make coupons for imported lobsters that were going on sale, the Chongqing Commercial Daily reported.

Tan, from coastal Guangdong province, said he wanted to sell four lobsters for 328 yuan, and the owner of the printery thought he meant 10.

“I didn’t know how much a lobster should cost. His Mandarin had a heavy accent. I carefully confirmed the quantity with him and asked if he meant 10, and he said ‘yes’,” the printery owner told the newspaper.

Tan, in turn, thought his counterpart said “four” instead of “ten”.

In the Chongqing dialect, “four” and “ten” are pronounced almost the same except their slightly differed tones – a delicate difference that many outsiders find difficult to distinguish.

The coupons were printed with “328 yuan for 10 lobsters” and were sold at the supermarket. The mistake was not found out because there were thousands of coupons for all kinds of products arriving on the same day, and the supermarket only ran a spot check on them, the report said.

Tan only discovered the mistake while he was doing the rounds at the supermarket the next day, by while time nearly 30 coupons had been sold.

The lobsters, imported from Boston in the United States, usually sell for 98 yuan each in the supermarket, and selling four for 328 yuan already left him very little profit, Tan said.

Selling ten for 328 had cost him more than 10,000 yuan, he added…..

From:  "Lost in translation: The difference between ‘four’ and ‘ten’ is more than 10,000 yuan for Chinese supermarket manager" (SCMP, 1/15/17)

VHM:  The title of the article is grossly misleading.  This has nothing to do with translation.

Transacting prices over the telephone like this, Tan was just asking for trouble.  But I don't think it was a matter of his Cantonese (if he indeed speaks that language) clashing with Chongqing patois.  Cantonese would be sei3 四 and sap6 四.  Those are too far away from MSM sì 四 and shí 十 and their Chongqing variants (discussed below) to cause a problem of misrecognition.  The Guangdong supermarket manager was probably trying to speak MSM, and either he did it relatively well or he may have spoken it with a Cantonese accent.

The problem is with the pronunciation of 四 and 十 in Chongqing topolect.  To tell the truth, I don't know exactly what it is, but — so far as I can recall from my memory of visits there about 15-20 years ago — it is similar to Chengdu topolect, both belonging to southwest Mandarin.

According to Tom Bishop, commenting on the Chongqing pronunciation of 四 and 十,

Aside from the tone they're identical and resemble "si" in Putonghua. I think the tones are low falling for 十 (陽平, 2nd tone, which includes old 入聲), and rising for 四 (去聲, 4th tone). I could be wrong.

Brendan O'Kane has this to say about the Chengdu pronunciation of 四 and 十:

Not familiar with Chongqing vernacular, but I remember hilarious confusion involving Chengdu speakers pronouncing 四十 as sísì when I visited in January 2005.

Compare that with sìshí 四十 ("forty") of MSM.

I discussed Sichuanese pronunciation in this comment from the following post, "The enigmatic language of the new Windows 8 ads" (5/14/13):

If you are a speaker of MSM (Modern Standard Mandarin) and you travel across the Mandarin-speaking areas of China, you will be astonished at the enormous variety of tonal configurations for words. My wife was from Shandong and grew up in Sichuan, so I got a liberal dose of the tonal patterns of both areas. I became so conditioned to them that, if I were upstairs and someone called who spoke Mandarin with Sichuan or Shandong tones, I could tell very quickly to which my wife downstairs was speaking, since — even though she spoke beautiful MSM — she would switch into the very different tone patterns of her interlocutor. I often said that Sichuanese tones were "upside down" in relation to MSM (though they are consistent within their own phonological system).

See also " Tones and the brain" (3/3/15).

[h.t. Mark Metcalf; thanks to Anwei Fang]


  1. tsts said,

    January 17, 2017 @ 10:14 am

    "The problem is with the pronunciation of 四 and 十 in Chongqing topolect."

    Yes, but I think there is also a Cantonese angle to this, in that Cantonese speakers tend to pronounce 10 as si not shi when speaking Mandarin. So then the only way to distinguish between 4 and 10 is the tone, and if that gets switched at the other end by Chongqing topolect, as Brendan pointed out, you get this misunderstanding.

  2. Victor Mair said,

    January 17, 2017 @ 11:09 am



    "The Guangdong supermarket manager was probably trying to speak MSM, and either he did it relatively well or he may have spoken it with a Cantonese accent."

  3. Ross Presser said,

    January 17, 2017 @ 11:12 am

    I was about to talk about how horrible it is to have numbers so easily confused, and how lucky I am not to speak any form of Sinitic language. Then I remembered "forty" vs "fourteen", which can be confused when I say them quickly. Still it seems strange to this monolingual that numbers as distinct as 4 and 10 would have such similar sounding words.

  4. John Rohsenow said,

    January 17, 2017 @ 11:16 am

    Remember the "tongue twister": 四十四石獅子…..?

  5. Victor Mair said,

    January 17, 2017 @ 11:24 am

    sìshísì shí shīzi 四十四石獅子 ("44 stone lions")

  6. Coby Lubliner said,

    January 17, 2017 @ 11:30 am

    In German zwei rhymes with drei and therefore, to avoid confusion, Germans sometimes say zwo for the former. I am surprised that Mandarin-speakers haven't come up with a comparable solution.

  7. Bob Ladd said,

    January 17, 2017 @ 11:46 am

    @Coby Lubliner: Yes, Dutch sometimes uses zeuven instead of zeven ('seven') to avoid confusion with negen ('nine'), and English radio communication (military and I think air traffic as well) uses niner instead of nine to avoid confusion with five. I think the problem with Chinese is that, as VHM put it, "tones, consonants and vowels start flying off in all directions". You need a relatively uniform and widely spoken standard language for a solution of this sort to work. The Chinese are working on that, of course, but it will be at least a generation or two before it's in place.

  8. cameron said,

    January 17, 2017 @ 11:58 am

    As commented above, such difficulties in being precise when referring to numbers are quite widespread. In English we often read out numerals to make ourselves clear. We might say "fourteen, one-four" or "forty, four-zero" in order to make clear whether we mean 14 or 40.

    Do speakers of Chinese languages never do that?

  9. markonsea said,

    January 17, 2017 @ 12:41 pm

    @ Ross Presser

    "Forty" vs "fourteen"?

    I once understood I had been told by some enquiry service (can't remember which) that it took "forty-five minutes" to get between terminals at Heathrow.

    I later realised I had actually been told "four to five minutes" …

  10. cliff arroyo said,

    January 17, 2017 @ 1:19 pm

    @Ross Presser

    "Then I remembered "forty" vs "fourteen", which can be confused when I say them quickly. "

    I guess your'e not American? For me 14 and 40 are different enough because the voicing in the latter (and many American say sevendy and ninedy as well possibly to avoid confusion with 17 and 19).

    15 and 50 as well as 16 and 60 remain troublesome though.

  11. Guy said,

    January 17, 2017 @ 2:32 pm

    The -teen numbers usually have the stress on the second syllable (except in cases, such as when counting, where contrastive emphasis is naturally placed on the first syllable), whereas -ty numbers have it on the first. Of course, this isn't always enough to avoid confusion (it's probably about as contrastive as a tone difference) but it helps a little.

  12. JK said,

    January 17, 2017 @ 3:09 pm

    Some places seem to have very different ways of saying "ten," in the northern Hunan Changde topolect they say "yi pao", which would make it very difficult to misunderstand for another number.

  13. Coby Lubliner said,

    January 17, 2017 @ 3:20 pm

    It seems to be mainly New Zealanders who pronounce the -teen ending in a way that makes it liable to confusion with -ty.

  14. PB said,

    January 17, 2017 @ 3:29 pm

    @Coby Lubliner: zwo used to be the feminine form of zwei, see for example Grimm's dictionary.

  15. Adrian said,

    January 17, 2017 @ 3:44 pm

    On Quora, Gabriel Chan writes: "In Mandarin (Chinese), there is a similar phenomenon, but for 1 (yi) and 7 (qi or ch'i), so one becomes yao when disambiguation is paramount."

    Is this correct?

  16. liuyao said,

    January 17, 2017 @ 4:13 pm


    Yes, yāo for 1 is sometimes written 幺, though one could argue it's simply an alternative pronunciation for 一. At least in Beijing, it's quite common, if not universal, to say yāo for yī when the numeral occurs in a string of 3 or more digits (phone numbers, bus routes, etc.)

    In more military speak, seven is pronounced guai 拐 (crook) and zero becomes dong 洞 (hole). I don't think four or ten have special readings.

  17. Michael Watts said,

    January 17, 2017 @ 4:41 pm

    I was just wondering about that lions tongue twister! It appears in the excellent novel Bridge of Birds as 四十四死石狮, translated "forty-four dead stone lions" (and supposedly a tongue twister during the Han dynasty, when I imagine those various characters would have had rather different pronunciations than they do today).

    Where does that tongue twister come from? Is it meant to be 古文? (I notice, for example, that "lions" appears without a measure word, despite the fact that the relevant measure word, 只 zhī, fits right in with the rest of the sequence in terms of modern pronunciation.) Is it in fact good 古文, and would it still have been a tongue twister in older pronunciation?

  18. Michael Watts said,

    January 17, 2017 @ 4:44 pm

    Adrian, as it was explained to me, there is essentially zero risk of confusing yī and qī when speaking face to face, but telephone static can make yī sound like qī.

    Saying yao for yi in long digit strings is also what I was taught to do in Shanghai.

  19. David Morris said,

    January 17, 2017 @ 4:56 pm

    ESL students very often pronounce thirteen/thirty etc very similarly, especially if their language doesn't have (as many) final consonants and/or has a more even syllable stress pattern than English. Sometimes the context is clear: 'How old are you?' 'Seven-tee'.

    By the way, the equivalent Sino-Korean numbers are sa (4) and ship (10), so the problem is unlikely to happen.

  20. empty said,

    January 17, 2017 @ 5:11 pm

    I remember (as an American visiting England about 40 years ago) mishearing "eighteen" as "eighty". I'm sure that I heard the stress as being on the first syllable.

  21. Brendan said,

    January 17, 2017 @ 6:24 pm

    I'm not so sure about "yao" as a disambiguator — it's only ever used in reference to the digit "1," and never when counting things, so it wouldn't be any use in distinguishing between "一個 yi ge" and "七個 qi ge." (And tone sandhi rules for 一 might also help disambiguate the two, though as the case of 四 and 十 in Chengdu and Chongqing shows, there's no guarantee of that.)

  22. C. Forrest Treaze said,

    January 17, 2017 @ 6:34 pm

    "To tell the truth, I don't know exactly what [the pronunciation of 四 and 十 in Chongqing topolect] is"

    Things like this are easy to find. The English Wiktionary usually has topolectal pronunciations of common morphemes like these, so if you trust those, it's a matter of looking up四#Pronunciation十#Pronunciation
    and you'll find Chengdu pronunciation is given as /sz̩²¹³/ "four" and /sz̩²¹/ "ten".
    You can then look at
    according to which we should expect Chongqing tones to be 214 for "four" and 21 for "ten".

  23. Rubrick said,

    January 17, 2017 @ 6:44 pm

    English is quite lucky in having ten quite phonetically distinct digits (which made number recognition one of the first and most accurate forms of computer speech recognition).

    On the other hand it drops the ball completely with pairs like "can/can't".

  24. Eidolon said,

    January 17, 2017 @ 7:20 pm

    "English is quite lucky in having ten quite phonetically distinct digits (which made number recognition one of the first and most accurate forms of computer speech recognition)."

    In my experience, the area where English drops the ball the most is near homophones in the names for the basic letters of the alphabet. Over the phone especially, it can be difficult to spell without resorting to tricks like 'C as in Cat,' 'E as in Elephant', etc.

  25. Jenny Chu said,

    January 17, 2017 @ 8:04 pm

    It seems amazing to me that such widely-used languages as MSM and English have "allowed" such basic ambiguity to persist for so long. Why haven't the speakers of these languages drifted into some practical disambiguation that eventually becomes standardized? I recall being taught in my very first linguistics class that we observe disambiguation coming up when there are homophones – the example provided was the collapse of the vowels of pin, pen and pan in southeastern American English, which ended up being replaced with stickpin, inkpen, and frypan.

    Does it imply that numbers are less important to us as humans than pens, pins, and pans? Perhaps we can all just try to get along with the "one, two, many" counting system :)

  26. Mark Mandel said,

    January 17, 2017 @ 8:13 pm

    One of the numbers I often need to say is 22, which in that context is normally ordinal, "twenty-second". But on the phone or in a noisy place, I've found, it's easily misheard as "twenty-seventh", and so I have come to say "twenty-two" instead.

  27. J.W. Brewer said,

    January 17, 2017 @ 9:01 pm

    "Zwoa" remains current in various German dialects. I guess I don't know whether it's limited to feminine nouns with those dialects sticking with "zwei" for masc+neut or can be used regardless of gender. Unfortunately, the first example I can think of off the top of my head (this song title from an early '80's rock band who often sang in Bavarian dialect rather than standard German) involves a noun (= "cigarette") which happens to be f.

  28. Gianluca said,

    January 18, 2017 @ 5:49 am

    In Italian we have the possible confusion between sessanta (sixty) and settanta (seventy) and derivates, especially in Northern dialects and accents where double consonants such as SS and TT tend to be pronounced as single consonants S (sesanta) and T (setanta). When one needs to avoid confusion among the two, one simply stresses the double consonants a lot, to the point of almost tripling them.

  29. Hans Adler said,

    January 18, 2017 @ 8:38 am

    @J.W. Brewer: I think your observation about Bavarian "zwoa" is probably due to an unrelated phenomenon. Bavarian often though not always has "oa" where standard German has "ei". E.g., Bavarians count "oans, zwoa, drei" instead of standard German "eins, zwei, drei" (one, two, three). (Also, one of the standard examples of funny Bavarian words is the word for squirrel tail: "Oachkatzerlschwoaf", which would be "Eichkatzerlschweif" in standard German, though one would normally use different words and say "Eichhörnchenschwanz" in standard German.)

    The apparent irregularity of "zwoa" for "zwei" but "drei" for "drei" occasionally gives rise to musings among Bavarians. If it isn't due to some general principle, one might speculate that Bavarians started using standard German "drei" to make it more distinctive; one might also speculate that this strong distinction in Bavarian contributes to the practice of using "zwo" for "zwei" in standard German.

  30. Abbas said,

    January 18, 2017 @ 9:01 am

    @Gianluca: exactly the same in Spanish sesenta (sixty) and setenta (seventy). Similar problem solved again by stressing consonants and slowing vocalization with the months junio (June) and julio (July).

  31. Christian Weisgerber said,

    January 18, 2017 @ 10:32 am

    @Hans Adler: Modern German ei represents a merger of two separate Middle High German vowels, ī and the diphthong ei < Gmc. *ai. In various dialects these have distinct reflexes. You can even see this pattern when comparing English and German cognates, e.g. minemein, but stonestein.

  32. Bob Ladd said,

    January 18, 2017 @ 10:57 am

    @Gianluca: For anyone who is hard of hearing, the sessanta/settanta problem is worse than some of the others discussed in this thread, because no amount of emphasis is going to make up for the fact that the acoustic cues to /s/ and /t/ are effectively identical for anyone whose hearing cuts out above 2 or 3 kHz. My late father-in-law used to use a different strategy when confronted with an ambiguous number, namely spelling them out as sei-zero or sette-zero.

  33. Neil Dolinger said,

    January 18, 2017 @ 1:00 pm

    In Mandarin, aside from sì 四 vs. shí 十, context usually clarifies meaning when words are homophones, but we still see an aversion to naming the fourth floor in PRC buildings due to the similarity of sì 四 'four' with sǐ 死 'die'. The elevator call buttons skip from '3' to '5'. I wonder whether buildings in Sichuan also skip over the tenth floor.

  34. Gianluca said,

    January 19, 2017 @ 3:33 am

    @Bob Ladd: extremely good point, I have never thought of the possible further problems for people with earing impairments. And yes indeed, the specifications sei-zero (six-zero) and sette-zero (seven-zero) are well widespread to avoid ambiguity.

    @Abbas: indeed. That's another similarity between Italian and Spanish.

  35. flow said,

    January 19, 2017 @ 4:09 am

    @Abbas—In German, June (Juni ['j:uni]) and July (Juli ['ju:li]) are annoyingly close together, so people either repeat dates using numbers ("erster sechster" / "erster sieb(en)ter" to clarify between "erster Juni" / "erster Juli") or they go and use alternative pronunciations for the months, viz. Juno [ju'no:] and Julei [ju'lai] (btw never written that way). Notice how both stress shifts and vowels change.

    As for the basic counting numbers, I've also heard 'five' "fünf" spoken as "fünneff" ['fynef]. And as was said earlier, I don't believe that "zwo" for "zwei" is due to Bavarian influence; rather, until approx. the 18th c, German used to have no less than *three* words for 'two', namely "zween" (masc.), "zwo" (fem.), and "zwei" (neut.), which all but vanished during the 19th c. "Zwo" to disambiguate "zwei" is (acc. to Herrmann Paul 1966) only due to the introduction of the telephone in the early 20th c. (but of course we're lacking living memory and live speech recordings for the time before that).

    I'm really surprised that so many speech communities have numbers that are so close together phonetically. I mean, "zwei" [tsvai] and "drei" [drai] are perceived as perilous by native speakers; how can one put such a distinction on tone alone? And not provide an easy way out, like "Ju-nooh (nicht Ju-laai)", or "no!, I said I can *not* come along"?

  36. Edwin Schmitt said,

    January 19, 2017 @ 5:43 am

    The points on wikipedia @C. Forrest Treaze mentioned are not quite accurate. In my experience the pronunciation for these words in Chongqing and Chengdu are closer to /sz̩13/ "four" and /sz̩31/ "ten" (sorry, for some reason I can't keep the superscripts). Jerry Norman also documents these tones for Chengdu in his classic book "Chinese" on page 196.

    @Neil Dolinger No, buildings in Chengdu do not skip the 10th floor. Occasionally buildings will skip the 14th and 24th…I don't remember noticing if the newer, taller complexes skip the 34th.


  37. Usually Dainichi said,

    January 20, 2017 @ 12:23 am

    The Germanic words for 4, which usually start with /f/, should really be starting with something like /wh/ according to Grimm's laws (kʷ>xʷ), since it comes from PIE *kʷetwṓr. I think one theory is that it was influenced by the starting consonant of 5.

    And I believe there's a similar theory about Russian 9 [ˈdʲevʲɪtʲ], which should really be something like [ˈnʲevʲɪtʲ] (compare Italic and Germanic versions), but at some point was influenced by 10, [ˈdʲesʲɪtʲ]. (Possibly in all Slavic languages, but this is the example I know.)

    So apparently the "natural linguistic forces" actually sometimes pull numbers closer instead of apart.

  38. C. Forrest Treaze said,

    January 20, 2017 @ 9:12 am

    @Edwin Schmitt,
    "In my experience the pronunciation for these words in Chongqing and Chengdu are closer to /sz̩13/ 'four' and /sz̩31/ 'ten'"
    Thanks. As for "four", the Wiktionary page I've linked to says: "In flowing speech, the fourth tone (213) becomes a low rising tone ˩˧ (13)." I assumed that the parties involved in the coupon problem, whatever varieties of Mandarin they were using, would have been careful not to use "flowing speech" pronunciation.

  39. Chris Kern said,

    January 20, 2017 @ 6:07 pm

    I feel like English is clear between FIF-ty and fif-TEEN but somehow confusion still results, so maybe the stresses aren't clear in everyone's dialect, or they don't come out as clear as we think they are in our head.

    Worse than that is can vs. can't; because many speakers don't pronounce the "t" in certain cases, it can be hard to tell the difference — there should be a difference because "can" usually has a different vowel from "can't" (at least in my speech) but confusion results in any case.

  40. Janus said,

    January 24, 2017 @ 12:39 pm

    The difference between and shí in Standard Mandarin is clear enough that it’s not a problem.

    Get rid of the retroflexes like most of the southern half of the country, and it’s a difference between and . Much more difficult, but still manageable in general.

    Go to Chongqing where they reverse the tones so that is and is … and it’s no wonder you’ll end up ruining yourself selling lobsters!

RSS feed for comments on this post