A Record of Spoken Chinese

06 Jan 2005

According to Xinhua, Chinese linguists are finishing up a huge database of spoken Chinese, and they’re going to use it as a basis for a new dictionary and grammar book of modern Chinese. This is good news! Using actual spoken speech as the source should produce a much more useful dictionary.

I’m a little disappointed that all the spoken Mandarin Chinese samples come from Beijing, though. Yes, I know Beijing is the standard, but wouldn’t real linguists want to get a larger sample? A descriptive sample for the whole country?

But then, maybe they had no choice in the matter. They probably had to act in the interests of the Chinese government. I have this bad habit of thinking of the interests of foreign students of Chinese, who won’t necessarily be living in Beijing when they go to China.

Via Language Log, which has more linguistic commentary on the issue.


John Pasden

John is a Shanghai-based linguist and entrepreneur, founder of AllSet Learning.


  1. As a native Chinese myself, I speak both Mandarin and Wu dialect. I think it is a good idea to start this language database, but there are thousands of dialects, I wonder how big it would be.

  2. John, don’t you think that the standardization of Chinese is in the best interest of foreign students in all parts of China? (Not to mention the best interest of Chinese folk.) I know that I am happy that the younger folk of China usually speak a more standardized putonghua (than many elders) that I can understand. Sure makes things easier for me.

    I like these posts you make about language stuff, even if the message boards are relatively silent.

  3. Even though I, too, wish that larger area were covered by the survey, I understand that, in the face of the existence of so many dialects, any study would have to semi-arbitrarily set some limits for sampling. The limits may be determined by theory (e.g., what’s the standard) as where as by practicality (e.g., cost of sampling). Limiting the data source to one or a few big cities makes the data collection affordable and more easily controllable (adding each city means an additional set of linguistic factors to consider). It would be a good thing (for someone else) to initiate an effort to collect a spoken dialect database for each of many other dialects, too.

  4. I read an interesting bit in a blog by a former student of John’s a while ago. She was wondering if there were different dialects of English. She was concerned at the thought of going somewhere in the English speaking word and not being able to communicate in the local idiom. The truth is she would only have a problem with accents.

    There are different dialects of English but the standard language is powerful and all English speakers understand it.

    My question is, what is Standard American English? John you use it your “about”? Is it based on the speech of any particular area of the USA?

  5. It would be good to be able to access not so much different dialects of Chinese, but different ‘accents’ of Mandarin. For example, when my Sichuan Uncles speak what they call ‘Mandarin’ it’s all ‘si’ instead of ‘shi’ and they mix up ‘l’ and ‘n’. If we could get used to hearing different accents of Mandarin, we wouldn’t get such a huge shock on discovering that in China not everyone sounds like Dashan. Chinese people believe that these accents just show that the person can’t speak ‘proper’ mandarin, but that’s like saying that Americans can’t speak proper English because they don’t speak it like the Queen of England.

    I also think its great when schools in China get English teachers from all over the world, since we all speak English slightly differently (esp those downunder).

  6. Unrelated, but I was just looking at your list of China Weblogs, and saw the comment:
    “Blogs with beige backgrounds are blocked in China.”

    My first thought was “what do the Chinese have against Beige??” until realising it was referring to your table. Duh!

  7. The thing is that not only is complete standardization of a language an impossible dream, but that one person speaking a completely standard version of a language is as well. Because language varies so much between individuals, throughout an area, and over time, standards can only really exist in theory (and maybe on TV). The larger the population, the more difficult it is to adhere to a standard. That’s why nonstandard Chinese is everywhere in China. How ridiculous would it be if we tried to apply a “standard English” to the whole world (including accent)?


    Alaric, I feel that it would be helpful to make students of Chinese aware of the real situation rather than schooling them to believe in some mythical entity that really only serves as an ideal that can never be fully realized. (That said, I’m NOT saying that a national standard is bad; it’s a necessary component of a society’s modernization, and it certainly improves communication.)

    E¨®in, SAE (Standard American English) is based on the American Midwest accent. It, too, exists fully only in theory, however.

    Kaili, I agree with you, although I wouldn’t want anyone to be taught to speak screwed-up Sichuanese Mandarin. hehe. Regional variations can be taught without going totally nonstandard like your Sichuanese example. For example, ÕâÀï vs Õâ¶ù£¬ optional ¶ù»¯ (»¨ vs »¨¶ù), etc.

    Realistically, though, in the classroom it’s not really worth the time to make these distinctions unless the class is already at a pretty advanced level. Still, in my Chinese 1 class, the teacher let us choose to write in traditional or simplified characters, and then we had to stick with it. Following that model, it would be kind of cool if each student had to choose a region, like Beijing or Taiwan or Xi’an or Shanghai, and then learn some of the features of that area’s Mandarin and try to use them in class.

  8. To answer E?’s query, if I may be so bold, in reference to Standard American English; modern American Standard English is based on Mid-Western American English usage without any specific locality bias. This standard evolved not from the academic community, but from radio and television production beginning in the 1930’s. Also, it probably should be understood as based on usuage by educated white males at that time.

    Modern technology, I believe, will eventually level out many of the dialectical differences, if what is occurring in the United States and Japan is any indication. So I would suspect basing the data base on a standard is probably not all that bad, except if one wants to have a real understanding of specific lanaguage usuage in any one area (at this time). The one concern I have is that Beijing usuage of Putonghua uses the retroflex ¶ù whereas South of that regional area does not (and I am not sure whether the retroflex will be standard in all of China).

    I think kaili has a very apt observation, everywhere I go Putonghua can have some rather heavy accents that may make it difficult to understand, especially if the speaker is older or less educated.

    I was attending a meeting in Vietnam a few years ago. All the attendees were Vietnamese except for myself and another white guy, he was an Aussie. He sat next to me, I suppose because he thought English speakers ought to stick it out together. But when he started to speak to me I did not understand him at all, his accent was so heavy for me. (He was actually a pretty good bloke, and we enjoyed one another’s company as in time I was able to make adjustments and understood much of his conversation, but it took a while).

  9. JFS touched on exactly what I was thinking when I first read about the database and John’s initial reaction of disappointment for the lack of diversity among the language samples. The fact that Beijing uses retroflex ¶ù way more than throughout the rest of China may be a problem for the study. It would be way more useful (and more correct?) if they took putonghua samples from Nanjing, Chongqing, and/or Shanghai as well.

    Actually, I think a lot more people speak the non-retroflex version of Mandarin than those that do, and I can’t see that changing anytime soon (though I’m not really in a position to authoritatively say anything like this).

  10. If this spoken Chinese database is to be used to support a dictionary, then there is no need to worry about the -er pronunciations, since the Chinese dictionaries have a system to treat such words as phrases so the root word remains intact in the written dictionary. Huir = Hua-er where Hua remains intact. Qir = Qian-er, Qian intact.

  11. Only somewhat relevant, but here’s a page on the topic of SAE (and the nonexistence of an “accent-free” American English):


    (via LanguageHat.com)

  12. Thanks for the link John, and I’m surprised that it’s based on the midwest. I originally assumed it would be Washington DC, like standard French is based on Paris and now I know standard Chinese is based on Beijing. On this one Americans are have bucked the trend.

  13. Maybe I should have said “Mandarin” there, not Chinese.

  14. I am not quite sure that I follow you, John. A language standard is not a representation of what people actually use, but rather a standard as to what an educated person ought to employ. Just as in engineering and science, a standard usually consists of a range of values, and so it will be in a language standard, what pronunciation, vocabulary, and grammar falls within some acceptable range will be counted as standard, and that outside the range non-standard.

    Gin, you are quite right about the dictionary forms, but I fear that standards are used for much more than dictionary items. Quite often they will be used as instructional forms also. As a foreign learner, I learned my Mandarin with the retroflex “er”, but living in the lower Yangtze valley, I find it more of a nuisance than a value, and somewhat of a learning experience in itself adapting to its non use.

  15. Anonymous Says: January 8, 2005 at 11:56 am

    To Eoin:

    Well, keep in mind that in the case of France and England, their capital cities have an outsized influence on their country (literally; Paris and London both dwarf in size by several times any other French or English city). As all of the major media and powerbrokers are congregated in the capital, it’s natural that the standard is set by the capital.

    In China’s case, it’s a totalitarian country, and the “standard” that is spoken by the news broadcasters and taught in schools is whatever is mandated by the central government, even if the accent sounds strange to most of the country. Note that the Nationalists did not emphasize the Beijing retroflex as much. If the Nationalists had held on to power, standard Mandarin would be much closer to the Mandarin spoken in the Yangtze delta.

    Neither of these cases apply to the US. It’s much more free market than China and not dominated to nearly the same extent by any one city (Note that in Germany, which also doesn’t have any one city that accounts for 1/6 of its population, the standard Hochdeutsch is certainly not the German spoken in Berlin; or even the dense Ruhr Valley). Midwestern speech came to be considered the standard in part due to the early high-voltage radio stations in Chicago (back in the early part of this century, you could listen to WGN from the East Coast to the Rockies), but also in part to the large number of people who found Midwestern English the most natural. Remember that the Midwest is vast, and due to western migration, pretty much all of the area to the west of the Midwest were populated by former Midwesterners. I had read that in the ’30’s, the accent in San Francisco was exactly the same as that found in Chicago.

    Ironically, with the Northern Cities Vowel Shift going on right now, in 20-30 years time, the English spoken here in Chicago will be quite distinct from what we consider standard American English, which is evidently the English spoken by broadcasters in 1930’s Chicago.

  16. Another thing:

    They speak with various non-rhotic accents in the cities of the Northeast. Considering that the vast majority of the country speaks rhotic English, New York, despite it’s size, wealth, and media influence, can not hope to set the standard in the US–it’s natives just sound too peculiar to the rest of the country. As for Washington, remember that it was for most of its history a sleepy government town that shut down completely when Congress took it’s summer break . It has as much chance for setting the language standard in the US (without resorting to fiat) as Canberra does in Australia.

  17. JFS,

    What are you responding to there?

    Your description of “what an educated person ought to employ” sounds like a prescriptivist approach that might be applied in France.

  18. John, I am not attempting to re-invent the prescriptive vs. descriptive debates of American liquists of the 1960s and 1970s. Much of that debate appears to me to be rather silly, anyway. I am just going by what the term standard. Standards are by definition prescriptive, it is what one attempts to emulate.

    There are a number of little stories that come to mind. For instance, back a few years, the planet Uranus had its 15 minutes of fame and was on all the major US networks. The issue at that time was how to pronounce Uranus. I forget, but it was either CBS or NBC that set the standard for pronouncing Uranus. They set a standard, they did not take a poll to determine how most Americans pronounced Uranus. I do not know, but I suspect they polled a number of academicans.

    Likewise, My mother was born on Whidbey Island, Washington State. Her mother was from Canada (Vancouver) and her father from Ohio. Her paternal grandfather emigrated to Washington State after the civil war, her paternal grandmother was Ohio German, did not speak a work of English. To cut to the short, her spoken language was Ohioan, and they at that time used the retroflex. So Washington was actually pronounce Warshington. But that was not standard and that pronounciation has gone away as the education system uses standards, not descriptors for much education. Likewise, my father was from Arizona, from a long line of early settlers. I learned to pronounce Tempe as Tampe (with a short a for the first vowel). But again that is non standard, and with the large influx of new settlers and standards of pronounciation in eduation, it is now pronounced tempe (with a short e for the first vowel).

    Again, in Japan in the 1960s regional dialects were quite common and heard everywhere. My last trip to Fukuoka I did not hear a word of Hakata spoken in public, just the national standard. I suspect the Chinese government is going to go in the same direction. Long ago when Qin Shi Huang Di consolidated the empire, among all the standardiations conducted, was the standardization of the written language (many of the kanji was written differently in each of the different states, and perhaps differently within a given state). Standardization did not hurt the language. Technology was not developed enough to do the same with the spoken language. But that is not the case now.

    In the United States we eschew descriptive grammars (it became part of modern liberal orthodoxy to do so). It appears to me that one of the consequences of that policy is that most literate and educated Americans do not have a very good grasp of the English grammar. My wife use to teach technical Japanese for the University of Washington to scientists, engineers, and managers. She found it very difficult, very frustrating, to expain comparative grammar points when the students do not have an understanding of the basis for that comparison.

    Again, I would suspect that if the Chinese government is establishing a database of spoken language, it is for use as a standard, that is, a prescription for individuals to emulate; not as a ethnolinquistic description of how Beijingers spoke as the beginning of the 21st century. I grant that the prescription may derive from description, but I suspect it will be prescriptive in the end.

  19. Anonymous Says: January 9, 2005 at 11:43 am

    To JFS:

    “Standardization did not hurt the language.”

    You could argue that, but when 25% or more of the preexisting vocabulary is outlawed from use, it’s hard to argue that you haven’t lost linguistic richness.

    As for Chinese following in the footsteps of Japanese, I don’t know how variable the Japanese accents are, but the Chinese dialects are as different from one another as the Romance languages are from each.

  20. To posted on January 9, 2005 at 11:43 am:

    I am not quite sure as to what you are referring to when you wrote that “25% or more of preexisting vocabulary is outlawed….” As to your remarks concerning the variability of the Chinese languages, I am cognizant of that fact. What I was alluding to was that modern technology is having a significant impact on pronounciation within language groups, and across language groups, as well as to the use of languages themselves. I do not believe that whether the affected language is a dialect or is a seperate language is really a significant factor; a much more important aspect would be the size of the language base itself. As an illustration, in Alaska, a number of Indian and Aleut languages have and are disappearing from use as the speakers convert to English. I personally believe this is happening at a rather rapid rate today is due to the influence of modern broadcasting. In addition to the commercial needs of the speakers, they are just bombarded with English programs on TV, etc. So English has become a much more convenient and usefull language than their native tongue. Here in China, the older generation of speakers are quite often prone to use non-standard Putonghua, sometimes quite affected by their native dialect or language. But I see the younger generation using very standard Putonghua constantly. In some areas the public broadcasting may be done in the native dialect or language; but in many other areas it will only be in standard Putonghua. My belief is that the standard will become the prevailing norm, especially if the population base of the dialect is relatively small or dispersed over a large area. In Guangdong it may not happen, Hongkong TV has local language programs, perhaps does the local stations there also, I do not know. But even there, you may see the local dialects begin to disappear and be replace by both standard Putonghua and a “standard” Guanddonghua.

    As for linguistic richness, I am not sure I buy into that ecological requirement. I see language as rather utilitarian. Besides the daily commerce in things, it will provide a richness of ideas and feelings, etc. But I do not know if all humans speak one language or a plethora of languages that richness of ideas or feelings will be any different except for the vaiety of sounds used, etc.

  21. Bit behind the times again, but back to my comment re learning to understand different accents of Mandarin, I definitely agree not to be specifically taught the different accents, but at least to be told about them or briefly introduced to a few.

    Our distance learning tapes were all Beijing accent and the tutors at our university must use a Beijing accent. Personally I feel its more realistic to just use any Chinese speaker to teach just like any English speaker can teach English (although I generally have to switch to an American or British accent to make myself understood to strangers in China, or Americans for that matter!). I think its because the Chinese have a much stronger idea of “correct” Mandarin than we do of English (esp as an NZer). When a northern Chinese hears a Guangdong person speak Mandarin they say “they have bad Mandarin” not “they have a Cantonese accent”. I try to speak ‘standard’ Mandarin but drop all the obviously Beijing things (except when I’m mocking them 🙂 Although this is hard when learning vocab orally, if I pick up words in Sichuan or in Beijing, I’ll generally pick up the accent until I bother to look the word up in the dictionary!

    Interesting stuff about the origins of American English. Apparently Kiwi English is based on a variety of lower class British dialect from the 1800s. Such as the use of ‘eh’ (we spell ‘aye’, pronounced like the ay in May, the shortened vowels (bread and bred, there and their, beer and bear, are the same to us) and the raised tone at the end of the sentence.

  22. To JFS:

    BTW, that was me before.

    Anyway, about the 25%, Qin ShiHuang outlawed and burned all of the books that existed before the Qin took control of everything. Since the Chu and other Warring States had their own vocabulary with unique characters, 25% (or more) of the (written) vocabulary was essentially erased from Chinese.

    Concerning the utilitarianism argument, I guess I simply disagree with that view. To me, language is inextricably tied with culture, and when a certain language/dialect is lost, a certain part of that culture is lost as well.
    For instance, you could argue that food is also purely utilitarian, and we could all get by on hamburgers, fries, and salads, but I think that it would be a loss if no one knew how to cook in the style of the Yangtze river delta any more.

    Personally, I’d actually prefer China adopting English as a putonghua rather than (or along side) Mandarin (as India has). English, at least, is equally difficult for all Chinese to learn (thus doesn’t disadvantage Chinese from a certain part of the country more than Chinese from another part) and is foreign enough that I think it has less chance of killing off the local dialects than Mandarin does.

  23. Richard:

    I underswtand your point, and I have no qualms about it, I just do not think it will have that much difference on the human experience whether we have one uniform language or a plethora of languages. Concerning Qin Shihuangdi, he did not eliminate 25% of the vocabulary, he standardized the orthography. That does not eliminate vocabulary, just the way we record it. It is as if all the Americans converted to the way the British world spells or vice versa. It may remove the exoticness of writing, but it does not eliminate vocabulary.

    kaili: I quite agree with you, but with some limitation. When my wife studied English, her father had learned English in the Brit fashion, her local teachers spoke Standard American English, but with a heavy Japanese accent, and they had two foreign teachers, one from South Carolina with a heavy Southern accent and one from New England with a heavy New England accent. It wss too much for them. The students held of student boycott, and the poor Southern women was removed from her teaching chores. Too much variety in the beginning is not conducive to learning a foreign language or so I suspect. My University had a variety of speakers, but their Putonghua was rather standard. Which mainly means some would have the retroflex and others would not. Although we also were taught a some, but not most, of vocabulary in the Beijing dialect; and even though I live and work in the Wu area, I tend to use nar instead of nali, etc.

  24. JFS, I guess you are right there! Actually the first time I taught English in China, the English teacher ‘translated’ everything I said for the students, that is, she repeated what I said in a Chinese accent.

    After reading your comment I thought about when I met my first American southerner, in China actually. It was amazing. I needed subtitles. Like “O brother where art thou?” (which I did use subtitles for).

  25. To JFS:

    Hard to say. We know that a bunch of characters were removed from usage. I’m not so certain that the characters that were removed all have a corresponding character in the Qin character set with the same meaning. What little I know of languages would hint against it.

  26. Richard:

    I see your point, but you may want to consider the relationship between the literary language and the local vulgate languages (there may not be a significant relationship). The languages I would suspect to be significantly divergent from the Qin language would be those spoken in the Yue, Wu, and Chu regions. But even here, I suspect the literary language used was similiar in all the regions, only the orthography being a variant. I could be wrong, but that is what I would suspect to be the case.

  27. Just wanted to know what would be the most efficient way to learn chinese? Nice posting am adding you to my fav list!

    Take care

Leave a Reply

Your email address will not be published. Required fields are marked *