读万卷书

Samuel Purchas died in 1626. He claimed never to have travelled more than 200 miles from his birthplace, in the East of England. In 1613 he published the first edition of Purchas his Pilgrimage. The title page of the fourth edition, published in the year of his death, explains the focus of Purchas’ work:

Relations of the world and the religions observed in all ages and places discovered, from the creation unto this present

Chap. 18, Section VI of that fourth edition offers us a single paragraph on Chinese language. His sources seem to be Matteo Ricci and Nicolas Trigault, to whom he admits, somewhat grudgingly, that he is indebted.

I would like to wish all of our readers a very happy and healthy 2013, and offer you Purchas’ account of language in China.

Note: What follows formed a single paragraph. I have modified spellings, except in the case of proper nouns, but retained the original punctuation marks.

The beginning of this discourse must be with their words, letters and writing; wherein this first to be admitted, that they have not one book written in the vulgar idiom or common language.

But they have one language called Quonhoa1, for their Courts, and writings, which is common through all China, which alone the Jesuits learned, and which the learned and strangers commonly use; women also and children attaining by this common use to the understanding thereof.

As for the differing languages of each Province, it is not necessary or commendable, being but of vulgar both use and reckoning.

But in every Tongue and Dialect the words are everyone Mono-syllables, however sometimes two or three vowels fall into one diphthong.

As for them, they mention not vowels or consonants, or letters, but in writing, the letter, syllable, and word is all one, being nothing else but hieroglyphical characters, of which there are no fewer than words, or things; which they yet so compound and connect, that they have not above 70, or 80 000.

If we pronounce any of their words in two syllables, it is when two of their characters are applied to signify one thing.

Some 10 000 of these characters are necessary for usual writing: for to know them all, is that which few either can, or need.

Their sound is also in great part the same, and yet both figure and signification different; so that there is no so equivocal language; neither can a Hearer write out an Oration or Speech from the Speaker’s mouth, not a book be understood of them which hear it read, but they must look, and discern with their eyes that equivocation which their ears cannot.

And in speaking they are often hereby forced sometimes to repeat that which hath before been elegantly delivered, sometimes to write it, or (if such means be wanting) with water on the Table, or Characters formed with the finger in the air, to express their minds to the conceit of others: and this is most common amongst the most learned which speak in print, and affect ink-horn rhetoric.

They have fine accents, by which they also distinguish this equivocation, that one and the same word thus by accents diversified, shall signify five several things nothing alike.

This makes the language hard to be learned of strangers; which yet the Jesuits learned to write and read: and I would all the Equivocators amongst them (that teach to illude oaths, and delude the World by their two-fold, two-forked, serpentine Equivocation in Mental reservations, and Verbal double-significations) were all there, learning the China language to convert Heathens, rather than practising the Romish equivocating Dialect to pervert Christians to worse than Heathenism.

Perverse Masters, lovers of strange language, in Prayers to God, in Oaths and Assertions to Man; in the one, Parrots without reason; in the other Devils, without Religion: this being the strongest bond which religion hath, binding at once to God and Man, and yet these Religious Mountebanks, by juggling quirks dissolving these bonds, and at once deluding both God and Man.

Foolish Romans! that sent back the Legates to Hannibal, that by equivocation had before fulfilled their Oath of returning! foolish Regulus! that returnedst to thy Tormentors, choosing they self rather than thy Oath to be tortured! and  most most foolish Martyrs, that so sleightly for want of this sleight, ran upon Fire, Swords, Lions!

And might not we begin a contention with that assertion, That an Oath for confirmation is to men an end of contention, which in this equivocating Hydra is rather multiplied?

That neither Rome Ethnic, nor primitive Christian Rome, could (at least by imitation of diabolical ambiguous Oracles) devise in those days so transcendent a subtlety, but Modern Rome by Jesuitical midwifery, must be the Mother of so super-fine a babe!

But what doth this Brat in our way?

I will rather follow the Jesuits in China than in Rome (except when Rome follows them thither too) and herein with thankfulness accept their report.

The reason of this equivocal sound of words is ascribed to the Chinois account of eloquence, in writing rather than speaking, and therefore to furnish that, neglecting this; insomuch as familiar messages are sent by writing, and not by word of mouth.

Musical skill was a good help to the Jesuits in learning the language, by reason of their variety of accents. And although this multitude of Characters be to the memory burdensome, yet it helps it as much as another way in saving the labour of learning diverse languages, while every Province of China speaking diversely, agree in writing; the Japonians, also Corayans, Cauchin-Chinois, Levhiees, all conceiving the same Characters, although the Japanese have an Alphabet of letters to write after our manner, which the Chinois have not.

They write their lines from the top of the Page to the bottom downwards, which they multiply from the right hand to the left: whereas our custom is quite contrary, from the left hand, sideways.

We have three consonants B.D.R. which the Chinois neither use, nor can by any Character express: and in our words which have them they borrow some sound nearest the same.

Likewise, they never have two consonants without a vowel between: and all their words end in vowels, except M. or N. of consonants only.

This and the diverse pronunciation of their Characters in diverse places, made the Latin form of Baptism hard to be expressed by the Jesuits.

1 Margin Note: Quonhoa signifies of the court, by this means the Magistrates need not in every new Province learn a new language.

35 responses to “读万卷书”

  1. Steve (Syz) says:

    As for the differing languages of each Province, it is not necessary or commendable, being but of vulgar both use and reckoning.

    Sigh. I think I heard a Beijinger say that yesterday…

    Quonhoa for 官话 is new to me, but pretty recognizable through the ages.

    Those numbers for “character knowledge” don’t seem that far off. Total of 70-80k. Educated folks know 10k.

    Great stuff!

    • Sima says:

      I thought you’d appreciate that, Steve.

      Some of the ‘pinyin’ used throughout is quite interesting, and probably worth a post in its own right. Purchas notes in the margin at one point that, of the capital, “Ricci calls it always not Paquin, but Pequin.”

      Things like the number of characters seem to be widely accepted. This third-hand description of the language – translated from Italian, through Latin to English – has quite a lot going for it. I wonder how many inferior descriptions of the Chinese language have been published in the past 400 years.

      Would anyone care to dispute any of what Purchas says about the language? (Accepting that he does get a little distracted with religious matters!) Is there anything he says that is just plain wrong?

      • Well, isn’t that wrong:
        “We have three consonants B.D.R. which the Chinois neither use, nor can by any Character express: and in our words which have them they borrow some sound nearest the same.

        Likewise, they never have two consonants without a vowel between: and all their words end in vowels, except M. or N. of consonants only.”

        • Sima says:

          Yes. Strange that it wasn’t G, D, B, though ‘R’ kind of deserves a special mention for all the trouble it’s caused over the years.

          Not quite sure how to view M and N, as I believe final M had been in use pre-Ming. Might it have been around when Ricci got here?

        • It’s not wrong looking at it like this: there is no voiced b or d (or g, but he skips that), and the English r is nothing like the Chinese r. Hanyu pinyin has the letters b and d (and g), but they represent unvoiced sounds; they lack aspiration, contrasting with p, t, and k, which are the aspirated versions.

          N is the only final consonant phoneme in current Mandarin. I have no idea if M ever was.

          • Sima says:

            Randy, are you claiming that the pinyin final ‘n’ constitutes a final consonant, but that the ‘ng’ doesn’t?

            Interesting that the lack of voiced plosives should prove such a concern, when to my ears the pairs b/p, d/t, and g/k ‘obviously’ mapped to their English counterparts.

          • Kellen says:

            M was once a final consonant but I’m not sure if it was at the time of Ricci’s writing.

          • “M” has changed into “n” at some point. However, I thought it was way before 17th century… approx. 14th century, not that late. Maybe Samuel was reading old manuscripts while in China?

          • Peter Nelson says:

            Regarding the “B.D.R.”, yeah, [b] [d] [g] would have been more accurate.

            On the topic of nasals, I’m not really sure what time period of guanhua is being talked about. I know Nanjing mandarin was the basis of guanhua during the Ming dynasty, but I was under the impression final -m had already merged at that point.

  2. Oops! No. I forgot about the ng! (I knew there would be something.)

  3. Zrv says:

    Bear in mind that these descriptions ultimately go back to Europeans speaking Romance languages like Portuguese and Italian. In these languages /r/ is a flap, and there is nothing like it in Mandarin (nor was there at the time of these writings). As other commenters above have noted, the voiced sounds /b/ and /d/ (which are much more clearly voiced in those European languages than they are in modern English) are indeed absent in Mandarin. (The omission of /g/ from the list puzzles me; it could be an oversight, or there may be an explanation I’m not aware of.) Finally, when it comes to nasal endings, most Romance languages lack an /ŋ/ phoneme altogether. This is why Guǎngdōng, both syllables of which end in /-ŋ/ in nearly every variety of Chinese, was romanized “Canton” by the Portuguese — /n/ was as close as they could get. I don’t know enough about the history of Romance languages to speak confidently about “m”, but I do know that in modern Portuguese orthography final “m” is simply a marker of vowel nasalization; so it does not in this description necessarily indicate an /-m/ ending in Mandarin. (Does Italian have words ending in -m? Maybe not.) In short, you can’t take the use of nasal letters at the end of words at face value when dealing with Southern European romance languages.

    An important lesson when looking at old transcriptions and romanization is not to fall into the easy trap of interpreting the letter values based on modern English. If you look at South Coblin’s research on Mandarin pronunciations based on European renderings, he is always careful to consider the letter values at the time of transcription in the native languages of the transcribers.

    • Great observation!
      However, Purchas was an English cleric, so the case of “m” still isn’t solved. And what about “r”? If it was Modern Mandarin “r” and Early Modern English “r” I think he would be able to see it’s similar. Was “r” in XVIIth century Chinese different in pronunciation?

  4. Sima says:

    Purchas was working from Matteo Ricci’s journals (written in his native Italian), which had been translated (into Latin) and expanded by Nicolas Trigault (Flemish). Purchas would have been working from this Latin, as far as I can tell.

    I’ll try to dig out some examples of their ‘pinyin’ which might shed light on their employment of ‘m’, but Zrv’s insight is very useful. Until I started reading this book, I don’t think it had occurred to me (I’m ashamed to say) that Canton was a phonetic spelling!

  5. Zrv says:

    Sima, you shouldn’t feel ashamed. The connection isn’t obvious because today Canton refers to Guangzhou, not Guangdong.

  6. Zrv says:

    Ryszard, as Sima notes, Purchas apparently never heard Chinese spoken and was working entirely off written descriptions. So it isn’t a matter of him noticing or not noticing similarities to English sounds.

    But your question about earlier Mandarin pronunciation is certainly relevant, for the questions both of ending -m and initial r-. I can now amend and clarify some of the things I said in my earlier post. The Korean alphabetic transcriptions of mid-15th-century Mandarin by Sin Sukju show that -m and -n were still distinct in the prestige variety of Mandarin that he was describing at that time. For example, he transcribes the word for ‘heart’ (心) as , contrasting with in other words like ‘new’ (新). However, by the time Ricci and Trigault were describing Chinese 150 years later around 1600 (based, apparently, on the prestigious pronunciation of Nanjing), -m had changed to -n. Trigault transcribes ‘heart’ as . But Trigault uses the letter -m to transcribe the ending [ŋ]! For example, he transcribes 經 as . This is no doubt related to the lack of -ng and -m endings in 17th-century Romance languages, and the use of -m to indicate nasalization (e.g. in French and Portuguese). But of course Purchas would not have been aware of this; he saw Trigault use the letter -m, and assumed that it represented an [m] sound.

    In Trigault’s transcription, the ancestor of modern Mandarin r- is transcribed as , implying a pronunciation like [ʐ] or [ʒ] (i.e. a voiced fricative similar to the value of the letter in modern French). Recall that in Wade-Giles transcription is also used for this sound. Some Mandarin speakers today still pronounce r- with heavy frication, more like [ʐ] than [ɹ], which can make it sound quite different from English [ɹ].

    You can find a good basic introduction to this transcriptional material in the works of W. South Coblin. A place to start is the first few pages of his 2000 article “A Brief History of Mandarin”, published in the Journal of the American Oriental Society 120.4:537-552, which is where the examples I’ve listed are taken from.

    The only remaining mystery, I think, is Purchas’s omission of voiced /g/ as one of the sounds lacking in Mandarin. It’s possible that this is because Trigault used the letter to represent some other sound, but I don’t see evidence of this in the few example transcriptions that I have in front of me at the moment.

    • I’m sorry, I’d like to understand your post in the fullest, however, it’s impossible since I don’t see the most substantial words!:
      “For example, he transcribes the word for ‘heart’ (心) as , contrasting with in other words like ‘new’ (新). However, by the time Ricci and Trigault were describing Chinese 150 years later around 1600 (based, apparently, on the prestigious pronunciation of Nanjing), -m had changed to -n. Trigault transcribes ‘heart’ as . But Trigault uses the letter -m to transcribe the ending [ŋ]! For example, he transcribes 經 as . ”

      I hope you’ll be able to provide missing parts but I already thank you for providing this insightful and extensive reply!

  7. Zrv says:

    Well, indeed, a Google books search does reveal that for some syllables (again based on the spelling conventions of Romance languages) Trigault employed instead of for the Mandarin r- sound. This is why Purchas did not mention /g/ as a missing sound in Mandarin. Check out the table of initials on page 50 of Dicionário Português-Chinês : 葡汉辞典 (Pu-Han cidian): Portuguese-Chinese dictionary*, under the initial /ʒ/ (ancestor of modern Mandarin r-), where you’ll see that was used before letters and while was used before letters and . Take a look too at the table of finals on the following pages. If you just scan the columns headed “RCS”, you’ll get a sense of what Purchas was looking at, without of course fully understanding the phonetic values represented by the letters.

    *http://books.google.com/books?id=A7h5YbM5M60C

  8. Zrv says:

    Hm, I think those dropped out because I had angle brackets around them, and they got interpreted as HTML tags. Trying again below with different formatting:
    —–
    For example, he transcribes the word for ‘heart’ (心) as “sim” (심), contrasting with “n” in other words like ‘new’ (新). However, by the time Ricci and Trigault were describing Chinese 150 years later around 1600 (based, apparently, on the prestigious pronunciation of Nanjing), -m had changed to -n. Trigault transcribes ‘heart’ as “sīn”. But Trigault uses the letter -m to transcribe the ending [ŋ]! For example, he transcribes 經 as “kīm”.
    —–
    From the above we can see that the two letters “M” and “N” mentioned by Purchas as the only ending consonant sounds in Mandarin, were in fact used by Ricci and Trigault to represent the sounds [ŋ] and [n], respectively.

  9. Zrv says:

    I see that I’m similarly missing elements from subsequent posts. Sorry for the confusion! Here are two more corrections:
    —-
    In Trigault’s transcription, the ancestor of modern Mandarin r- is transcribed as “j”, implying a pronunciation like [ʐ] or [ʒ] (i.e. a voiced fricative similar to the value of the letter in modern French). Recall that in Wade-Giles transcription “j” is also used for this sound. Some Mandarin speakers today still pronounce r- with heavy frication, more like [ʐ] than [ɹ], which can make it sound quite different from English [ɹ].
    —-
    Well, indeed, a Google books search does reveal that for some syllables (again based on the spelling conventions of Romance languages) Trigault employed “g” instead of “j” for the Mandarin r- sound. This is why Purchas did not mention /g/ as a missing sound in Mandarin. Check out the table of initials on page 50 of Dicionário Português-Chinês : 葡汉辞典 (Pu-Han cidian): Portuguese-Chinese dictionary*, under the initial /ʒ/ (ancestor of modern Mandarin r-), where you’ll see that “g” was used before letters “i” and “e” while “j” was used before letters and “o” and “u”. Take a look too at the table of finals on the following pages. If you just scan the columns headed “RCS”, you’ll get a sense of what Purchas was looking at, although he did not of course fully understand the phonetic values represented by the letters.

    *http://books.google.com/books?id=A7h5YbM5M60C

  10. Zrv says:

    Okay, this is all a bit of a mess, so I’m going to try to put everything together into one coherent post without any missing elements.

    ——
    But your question about earlier Mandarin pronunciation is certainly relevant, for the questions both of ending -m and initial r-. I can now amend and clarify some of the things I said in my earlier post. The Korean alphabetic transcriptions of mid-15th-century Mandarin by Sin Sukju show that -m and -n were still distinct in the prestige variety of Mandarin that he was describing at that time. For example, he transcribes the word for ‘heart’ (心) as “sim” (심), contrasting with an “n” ending in other words like ‘new’ (新). However, by the time Ricci and Trigault were describing Chinese 150 years later around 1600 (based, apparently, on the prestigious pronunciation of Nanjing), -m had changed to -n. Trigault transcribes ‘heart’ as “sīn”. But Trigault uses the letter -m to transcribe the ending [ŋ]! For example, he transcribes 經 as “kīm”. This is no doubt related to the lack of -ng and -m endings in 17th-century Romance languages, and the convention of using -m to indicate nasalization (e.g. in French and Portuguese). But of course Purchas would not have been aware of this; he saw Trigault use the letter “m”, and assumed that it represented an [m] sound.

    In Trigault’s transcription, the ancestor of modern Mandarin r- is transcribed as “j”, implying a pronunciation like [ʐ] or [ʒ] (i.e. a voiced fricative similar to the value of the letter in modern French). Recall that in Wade-Giles transcription “j” is also used for this sound. Some Mandarin speakers today still pronounce r- with heavy frication, more like [ʐ] than [ɹ], which can make it sound quite different from English [ɹ].

    You can find a good basic introduction to this transcriptional material in the works of W. South Coblin. A place to start is the first few pages of his 2000 article “A Brief History of Mandarin”, published in the Journal of the American Oriental Society 120.4:537-552, which is where the examples I’ve listed are taken from.

    Update to the above post, on “g”:
    As for Purchas’ failure to mention /g/ as a voiced sound absent from Chinese, this is because the letter “g” is in fact present in the Ricci/Trigault transcription system. For some syllables (again based on the spelling conventions of Romance languages) Trigault employed “g” instead of “j” for the Mandarin r- sound. This is why Purchas did not mention /g/ as a missing sound in Mandarin. Check out the table of initials on page 50 of Dicionário Português-Chinês : 葡汉辞典 (Pu-Han cidian): Portuguese-Chinese dictionary*, under the initial /ʒ/ (ancestor of modern Mandarin r-), where you’ll see that “g” was used before letters “i” and “e” while “j” was used before letters and “o” and “u”. Take a look too at the table of finals on the following pages. If you just scan the columns headed “RCS”, you’ll get a sense of what Purchas was looking at, although he did not of course fully understand the phonetic values represented by the letters.

    *http://books.google.com/books?id=A7h5YbM5M60C

  11. Sima says:

    Zrv, 辛苦了!
    That all seems to fit with what I’ve seen in Purchas.

    Can you recommend any good sources for information on MSM initial ‘r’?

  12. Zrv says:

    I’m not aware of any studies specifically of Mandarin initial r-. Are you thinking about phonetic studies of synchronic variation, or historical developments from older nasals?

    • Kellen says:

      I’m going to answer for him because it’s part of a larger discussion: phonetic studies of synchronic variation.

      • Sima says:

        Yes. Thanks, Kellen. I’m interested to know more about the current position with regard to ‘standard’ pronunciation of ‘r’, common variations from that standard, and the possibility that pinyin ‘r’ covers more than one kind of articulation (notably, possible differences between ‘ri’ and other ‘r’-initial syllables.

        Zrv, your use of [ʐ] and [ɹ] caught my eye, particularly the latter. It has been my perception that many speakers produce the ‘r’ in rang, ran etc with something much closer to the latter, but wonder whether you would use [ɻ].

  13. Sima says:

    @zrv
    At the risk of going off in another direction altogether…
    “In Trigault’s transcription, the ancestor of modern Mandarin r- is transcribed as “j”, implying a pronunciation like [ʐ] or [ʒ] ”

    It’s interesting that ‘j’ is so flexible/variable. Purchas mixes the letters ‘i’ and ‘j’:
    ‘subiects’
    ‘Presbyter Iohn’
    As far as I can tell this was the norm at the time. I’m not suggesting that this bears particularly on his use of Trigault’s transcriptions.

    In some European languages, the letter ‘j’ has (or had) a value close to modern English ‘y’ /j/. Can we be sure that multi-lingual Ricci and Trigault would have used it with /ʒ/ in mind?

    Somehow, the transcription of 人 as ‘jen’ always offended me, particularly when I heard non-natives pronounce it as /ʒen/, but when I found that many people around me (in NE China) mix up pinyin ‘r’ and ‘y’ (样 and 让 are frequently both pronounced as the former), and 日本 becomes ‘yiben’, that really blew my mind.

    Actually, that ‘slip’ from ‘r’ to ‘y’ is very comfortable and natural. Some ‘awkward’ words like 软 (ruan3) in joined up speech seem to almost require it.

    • Kellen says:

      Somehow, the transcription of 人 as ‘jen’ always offended me, particularly when I heard non-natives pronounce it as /ʒen/

      I mentioned this before, but not on the site: This /ʒen/ aversion would have surprised me up until about a week ago. I hear native speakers say something close to /ʒen/, at least in the greater Shanghai area, though in some cases it morphs into something closer /zen/. I really had no idea Northeasterners (Beijing/Hebei included here) used [ɹ] or [ɻ] for r-initials.

    • “In some European languages, the letter ‘j’ has (or had) a value close to modern English ‘y’ /j/. Can we be sure that multi-lingual Ricci and Trigault would have used it with /ʒ/ in mind?”

      Polish “j” sounds like English “y” in ex. you. :) Latin “j” also sounds like “y”.
      I pronounce Chinese “r” as /ʐ/, though I’m not sure if it’s good. 😛

  14. Zrv says:

    Trigault’s book Xīrú ěrmù zī 西儒耳目資 was intended to teach Westerns Chinese pronunciation and characters; the Romanization is accompanied by explanations of pronunciation and the values of the letters. I don’t think there is any doubt among people who have studied these materials that letter “j” was intended to represent a voiced fricative of some sort.

  15. Zrv says:

    On the broader questions of the history of this sound, the source is a Middle Chinese palatal nasal, which we could notate as [ɲ] or [ȵ], similar to Spanish “ñ”. It has developed in many different directions in different dialects. It is realized as zero or y- (i.e. [j]) in Cantonese and Sino-Korean and some Northeastern dialects; as a fricative z- in some Wu and Min dialects; as affricate “j” (i.e. [dʒ]) in one of the Sino-Japanese layers (cf. Nihonjin 日本人), as [n] in the other (cf. Nihonjin 日本人 again!), and of course as [ɹ] or pinyin er in Standard Mandarin. Those are just some of the variants. Karlgren reconstructed the Middle Chinese initial “ńź”, a palatalized nasal with voiced frication, to account for all these developments, but there’s no need to posit anything other than a simple palatal nasal, from which all these developments are quite natural.

  16. Katie says:

    “A palatalized nasal with voiced frication” … wow, those historical linguists amaze me sometimes. Is that even physiologically possible?

    Thanks Zrv, really interesting posts. The ‘m’ ending would actually make a lot of sense for modern Mandarin, where vowels are nasalized before ‘ng’. The nasalization is probably more prominent than the consonant to a person who speaks a language like Portuguese or French.

    I’ll also note along with Sima that you’re using [ɹ] for Standard Mandarin. How precise of a transcription would you consider that? Inquiring minds want to know. (Sorry, by offering a transcription for the sound you’ve unwittingly been dragged into a painfully long discussion on how to transcribe it.)

  17. Zrv says:

    Sima, Katie, sorry for the delayed response to your question about the phonetic realization [ɹ] for Standard Mandarin r-, as opposed to [ɻ]. I’m not a phonetician, but I don’t think I’ve ever heard anybody pronounce this initial as a true retroflex [ɻ], with the tongue curled back so that the underside of the tongue is facing up toward the roof of the mouth. If I make an [ɻ] sound, it does not sound like Mandarin r- at all. Regardless of the degree of frication, it seems to me that a Mandarin r- is always pronounced with the body of the tongue retracted and the blade bent back, but the tongue tip in the normal orientation and approaching the palate just behind the alveolar ridge. That is precisely an [ɹ], though not exactly the same as English [ɹ], which usually has a bunched tongue body.

    I have never seen Standard Mandarin r- notated as [ɻ] even in close phonetic description. (If you know of a published source with this symbol, do let me know.) Norman in his book “Chinese” notates [ɹ] and so does Yen-Hwei Lin in “The Sounds of Chinese” (2007) (see page 284 and the description on page 28 of the sound as an “apical post-alveolar”, not a retroflex).

    If r- isn’t a retroflex, this naturally raises the question of why increased frication would lead to the notation of [ʐ] instead of [ʒ]. The answer is complicated. As it turns out, NONE of the Mandarin consonants are true retroflexes, as has been demonstrated definitively by the Hong Kong linguist Eric Zee. That is to say, in the pronunciation of zh-, ch-, and sh-, the underside of the tongue blade does not make contact with the roof of the mouth as it does with the true retroflexes found in India. Karlgren, I think, takes the original blame for mistaking the Mandarin post-alveolar apicals sibilants for true retroflexes.

    But here’s the funny thing: everybody still notates them with retroflex symbols [tʂ tʂʰ ʂ], even those who know better, and the reason for that, I think, is that IPA lacks convenient accurate symbols for those sounds.

    • Kellen says:

      We’ve used the Zee&Lee paper in earlier stages of the grand discussion. I’m familiar with Yen Hwei Lin for work on Huojia but hadn’t taken a look at The Sounds of Chinese. Reading some sections now, she’s got some good points, especially regarding j/q/x.

      I’ll say for myself at least that one of the greatest frustrations with IPA is it’s complete failure to do what, in my mind, it should have been designed for in the first place. Like much of the field of linguistics, it feels very much originally intended for stock Western languages.

      • Sima says:

        Zev,

        Good of you to come back to us on this…

        If r- isn’t a retroflex, this naturally raises the question of why increased frication would lead to the notation of [ʐ] instead of [ʒ]. The answer is complicated. As it turns out, NONE of the Mandarin consonants are true retroflexes, as has been demonstrated definitively by the Hong Kong linguist Eric Zee.

        Oh why, oh why couldn’t someone have told me this at the beginning instead of leaving me to work it out for myself!

        IPA very far from helpful here. But given that we’ve pretty much resigned ourselves to using retroflex characters for ‘zhi, chi, shi,’ I’m leaning toward [ɻ] as a wholly reasonable account of ‘r’. [ʐ] has way too much friction, [ɹ] just doesn’t get close for me.

        The one element that really impacts here is whether there’s any vowel (or vowel-like thing). Do you have a view, or would you recommend a view on this, Zrv?

        Kellen, Katie…Zee & Lee paper? Any chance of a link?

  18. Zrv says:

    I agree, if we are going to use retroflex symbols [tʂ tʂʰ ʂ] to represent the not-quite-truly-retroflex zh ch sh sibilants of Mandarin, then we might as well also use the retroflex symbol [ɻ] to represent the not-quite-truly-retroflex r liquid. It’s easy to complain about IPA, but the fact is that the basic symbols of IPA are not suited to, and are not designed to, enable close transcription. There are too many subtle variations in phonetics across languages, and IPA would just be unworkable if it was designed to do this. So to do close transcription, you have to modify the basic letters with diacritics. So you’ve always got this trade-off in your IPA notation: phonetic accuracy vs. visual clutter.

    I take it your question about vowels has to do with the syllables zhi chi shi ri and zi ci si? There are two ways to approach the question, one is phonetic and one is phonological. Then answer to the phonetic question depends on how you define “vowel” phonetically; the answer to the phonological question depends on how you conceived of the Mandarin phonological system and your position on some general theoretical questions in phonology. So for me, I don’t think there is any one definitive answer to the question, and in some ways it’s a meaningless question.

    Phonetically, if you define a vowel as an articulation in which there is no constriction of airflow in the vocal tract, then there are clearly no vowels in any of these syllables. Airflow is clearly constricted, which is why we hear a “buzzing” or fricated quality in the syllables. On the other hand, if you simply define a vowel as any sound that can carry the suprasegmental features that a syllable needs (pitch, length, loudness, etc.) then you could argue that these are vowels regardless of how constricted they are.

    Phonologically, it’s much more complicated. First you have to decide if your theory of phonology requires a phonological vowel in every syllable. If so, then you could argue that these are consonantal sounds which, by virtue of being in the vowel slot of the syllables, are functioning as vowels in terms of syllable structure. This is the same kind of argument that says that the [n] sound in English “button” is a vowel. It does all the work of a vowel: carries pitch and stress and length. You could take a different tack and argue that these sounds are conditioned allophones of the vowel /i/ which are phonetically consonants. But because they are part of a vowel phoneme, we treat them as vowels.

    If your theory of phonology does not require that syllables underlyingly have vowels, then you could argue that these syllables all have zero-vowels, i.e. that the vowel slot is empty. (This is the analysis that underlies the bo-po-mo-fo rendering of these syllables.) Because surface realization of a consonant has to have a sound to carry pitch, tone, and length, the initial consonant sound simply spreads into the empty vowel slot to fulfill this function.

    All of these different analyses “work” — they all allow you to devise a consistent phonological structure for Mandarin syllables and to derive surface phonetic forms using a small set of rules. So you can choose whichever analysis you like — being aware that some of those analyses contain implications for how the phonology of other languages would be interpreted.

Leave a Reply to Randy Alexander