Phonetic annotations as a welcome aid for learning how to read and write Sinographs

« previous post | next post »

In several recent posts, we've been discussing the most efficient, least painful way to acquire facility with hanzi / kanji / hanja 漢字 ("Sinographs; Chinese characters").  Lord knows there are endless numbers of them and they are so intricately constructed that it is an arduous task to master the two thousand or so that are necessary for basic literacy.

It would be so much easier to learn the Sinographs if language pedagogues would provide phonetic annotations for each character.  Better yet, the phonetic annotations should be divided into words with spaces between them according to the official orthographic rules.

I've pleaded with Chinese language and education authorities to provide massive amounts of such reading materials, but so far my words have fallen on deaf ears.  As a matter of fact, two and more decades ago, there were plentiful amounts of such materials available in schools and bookstores, but under the new retrograde regime, things have gone downward and backward in the last few decades.

As I have mentioned in numerous posts (a few are cited below), I learned most of my written Chinese (reading and writing) — relatively painlessly — with the aid of Zhùyīn fúhào 注音符號 (bopomo ㄅㄆㄇㄈ ["Mandarin Phonetic Symbols'"]).  When I was living in Taiwan in 1970-72, I had the pleasurable convenience of reading local and international news, Chinese and world literature, history, recipe books, etc. in texts that were provided with bopomo phonetic annotations.  Similarly, in learning Japanese, I relied heavily on texts that were provided with furigana ruby phonetic annotations.  I am eternally grateful to the language educators who created such materials.  On the other hand, I believe that it is perverse for the current crop of language mavens to eschew such aids for language learners.

For the countless legions of those who suffer from character amnesia (mentioned in dozens of Language Log posts, a selection of which are listed below) and aphantasia (a seldom addressed, but equally challenging impediment to command of Chinese characters) — and with the advent of writing characters via computers (especially using Romanized or other phonetic inputting) and the gradual demise of handwriting, that includes an ever growing proportion of the population — it would be wise, kind, and humane to supply phonetic annotations for hanzi / kanji / hanja 漢字 ("Sinographs; Chinese characters"), at least for learners up to the acquisition of basic literacy (around 2,000 characters), and for difficult, rare characters even for literate readers.

Today in one of my classes we were reading Sun Yat-sen's (1866-1925, founder of the Republic of China) letter to Li Hongzhang (1823-1901, a high-ranking Qing Dynasty official), and we encountered this term:  頡頏.  None of the students (most were M.A. candidates from the PRC) knew how to pronounce the characters or what they meant.  This lexeme can mean "the flight of a bird up and down" or "equally matched") and is pronounced xiéháng / ㄒㄧㄝˊ ㄏㄤˊ.  It is also written and pronounced thus:  jiékàng 拮抗.  In my estimation, scholars, teachers, editors, and publishers should provide phonetic annotations for all such terms.  That would make it so much easier for readers to know how to pronounce them and to look them up by sound in dictionaries and databases.  Why not be as user-friendly and unintimidating as possible?  Other than to stroke one's own ego, what's the point of being user-unfriendly and intimidating?

 

Selected readings



26 Comments

  1. Philip Taylor said,

    April 26, 2019 @ 9:04 am

    My own introduction to Mandarin / Putonghua was courtesy of three teachers from SISU ("Shanghai International Studies University"), each of whom visited my home university in consecutive years. Each used Kan Qian's excellent Colloquial Chinese as the standard text through which to teach us (all Westerners) spoken MSM. The text (and accompanying electronic edition) makes great use of Hanyu Pinyin and hanzi are relegated to background study. However, several of us felt that Hanyu Pinyin, whilst infinitely more accessible than hanzi to a Western beginner, was nonetheless a very broad transcription (by comparing the written Pinyin with the spoken texts that accompanied the course), and we sought in vain for a dictionary that would use the IPA to provide a more accurate guide to pronunciation. Not only were we unable to find such a dictionary in the UK, even with the help of our Chinese teachers we were unable to find such a dictionary in Shanghai.

    Why is it, I wonder, that the IPA is almost universally used to provide an accurate guide to the pronunciation of the majority of the world's languages, yet is almost universally eschewed for MSM ?

  2. Antonio L Banderas said,

    April 26, 2019 @ 9:44 am

    @Victor Mair
    Ever since I read DeFrancis' classic "Visible Speech", I've wondered whether the figures ever held true about Chinese "phoneticity" up to 90% for "the two thousand or so that are necessary for basic literacy".

    @Philip Taylor
    There isn't any such resource either for other languages, namely showing a narrow IPA transcription of phonological processes. Otherwise, what are the shortcomings of Pinyin you are referring to?

  3. Philip Taylor said,

    April 26, 2019 @ 10:17 am

    Antonio — One general observation, and one specific example :

    General observation : in Colloquial Chinese, the pronunciation of Hanyu Pinyin is expressed in terms of English phonemes, with occasional further guidance when the author feels that the English phoneme is sufficiently far from the intended Chinese phoneme to justify this (e.g., "ao like ow in how", "r like r in road, but with the tongue loosely rolled in the middle of the mouth"). If one were to follow this pronunciation guide in isolation, one would be led to believe that (modulo tones), nĭ hăo should sound just like "knee how". For me, the sound that I make when I pronounce nĭ hăo, whilst vaguely similar to the sound of "knee how", is most definitely not the same, both elements differing to a greater or lesser extent.

    Specific example : the r of rènshi nĭ, as spoken on the accompanying audio materials, is totally unlike the r of rén in nà liǎng gè rén, and indeed the sound of the first is totally unlike the guidance given as to the pronunciation of r, yet both sounds are represented by the same Hanyu Pinyin symbol.

  4. Jonathan Smith said,

    April 26, 2019 @ 10:58 am

    @ Philip Taylor, you are right that for the persnickety knee is really not like ni, etc., but this is not a problem IPA can resolve — hence joking reference to the "International Phonemic Alphabet" (first seen on LL at some point…)

    Your observation re: onset r- is also interesting but hard to address without audio files… clearly realization of this phoneme differs dramatically from place to place and person to person… my view is that almost all casual descriptions one encounters are wrong; for one — he asserts without searching for specific articulatory phonetic evidence — r (and zh/ch/sh) are not "retroflex", an articulation totally foreign to Chinese as far as onset is concerned.

  5. Philip Taylor said,

    April 26, 2019 @ 2:30 pm

    [OT] "for the persnickety …" — I've seen "persnickety" used here before, but in <Br.E> we know it as "pernickety". Is "persnickety" (a) <Am.E>, (b) humo[u]rous, or (c) both ?

    On-topic : I will try to extract the audio segments representing "Rènshi nĭ, wŏ hĕn gāo xìng" and "Nà liǎng gè rén, shì shéi ?" and post links.

  6. Victor Mair said,

    April 26, 2019 @ 2:53 pm

    This might help:

    =====

    In phonology, an allophone (/ˈæləfoʊn/; from the Greek ἄλλος, állos, "other" and φωνή, phōnē, "voice, sound") is one of a set of multiple possible spoken sounds, or phones, or signs used to pronounce a single phoneme in a particular language. For example, in English, [t] (as in stop [stɔp]) and the aspirated form [tʰ] (as in top [ˈtʰɔp]) are allophones for the phoneme /t/, while these two are considered to be different phonemes in some languages such as Thai and Hindi. On the other hand, in Spanish, [d] (as in dolor [doˈloɾ]) and [ð] (as in nada [ˈnaða]) are allophones for the phoneme /d/, while these two are considered to be different phonemes in English.

    =====

    Source

  7. Michael Watts said,

    April 26, 2019 @ 3:29 pm

    I've been trying to teach Chinese to my brother.

    There is in fact a very large amount of material with all the characters phonetically annotated. I have this book [comment with link was swallowed; ISBN 978-7535386045], which annotates every character, covers subject material I enjoy, and is beautifully illustrated. But I barely read it myself because of the large amount of effort involved — this is a native book targeted at Chinese children who are assumed to already know the language.

    Phonetic annotation does nothing for people who don't know what the words mean. All it does is let me use pinyin input to look up dictionary entries instead of graphical input. With my brother, I read the Rainbow Bridge graded readers (for example [link was swallowed; ISBN 978-7513811927]), and they are much better suited to the task despite not including pinyin annotation, because they are written to a limited target vocabulary and generally stick to it.

    it would be wise, kind, and humane to supply phonetic annotations for hanzi / kanji / hanja 漢字 ("Sinographs; Chinese characters"), at least for learners up to the acquisition of basic literacy (around 2,000 characters), and for difficult, rare characters even for literate readers.

    This is already done for difficult, rare characters. I've seen it in dictionary entries and on plaques in the Forbidden City.

    the r of rènshi nĭ, as spoken on the accompanying audio materials, is totally unlike the r of rén in nà liǎng gè rén, and indeed the sound of the first is totally unlike the guidance given as to the pronunciation of r, yet both sounds are represented by the same Hanyu Pinyin symbol.

    The pinyin r is conceptually the voiced equivalent of pinyin sh. In English, the postalveolar approximant /ɹ/ ("r") and the postalveolar voiced fricative /ʒ/ (as in "vision") are phonemically distinct; in Mandarin Chinese they are not. On the fundamentals, they're both sounds that (1) are voiced and (2) position the tongue in the same place; the difference is in precisely how far the tongue is from the roof of the mouth. Anywhere along the continuum is fine for Mandarin.

  8. Philip Taylor said,

    April 26, 2019 @ 4:29 pm

    Jonathan, Victor, Michael — yes, "allophones" was basically what our Chinese lecturers said when we queried the difference between the sound of the initial r of rènshi compared to the initial r of rén, but the son of teacher number 2 offered a very different explanation. As he pointed out, despite the fact that rèn (as in rènshi ("to know") and rén (person) are written the same in Pinyin (modulo tone), in Hanzi they are written as two different characters (认 v. 人), and even though the first is clearly based on / derived from the second, there is no a priori reason why they should sound (exactly) the same. It would be interesting to know whether, despite the fact that there exist various allophones all of which can be represented in Pinyin as r, there is nonetheless a general tendency amongst native MSM speakers to produce more of a /ʒ/ sound for the initial r of rènshi compared to more of a /r/ sound for the initial r of rén.

  9. Jonathan Smith said,

    April 26, 2019 @ 4:44 pm

    @ "[t]he pinyin r is conceptually the voiced equivalent of pinyin sh."

    I guess it depends what you mean by "conceptually". It's certainly true that this is a convenient way of fitting Mand. /r-/ into a phonemic framework, and it is also "close enough for jazz" for non-native learners (likewise [ɹ] [ʒ] etc.) given that native listeners will not be troubled. But I don't think it's correct from an articulatory standpoint, which may be of theoretical/historical/etc. interest and matters for the persnickety (which yes is apparently a North American quirk.)

  10. Jonathan Smith said,

    April 26, 2019 @ 4:51 pm

    @ Philip Taylor, yes different words but of course a minimal pair on tone, so unless there is a claim for allophonic separation from tone difference (not impossible), I am very skeptical one could generate experimental support for the idea that "native MSM speakers […] produce more of a /ʒ/ sound for the initial r of rènshi compared to more of a /r/ sound for the initial r of rén." I.e., I would expect your observation to relate to the specific recording, to different speakers, to articulatory effort, to post-pausal position… or some such.

  11. David Marjanović said,

    April 26, 2019 @ 8:37 pm

    for one — he asserts without searching for specific articulatory phonetic evidence — r (and zh/ch/sh) are not "retroflex", an articulation totally foreign to Chinese as far as onset is concerned.

    "Retroflex" can mean some or all of:

    1) ordinary laminal coronal sounds, except so far postalveolar that the tongue feels like it's bent 90° upwards. These occur in Russian and MSM.
    2) the apical version of 1): the tongue points straight upwards, and its very tip articulates with the palate.
    3) the tongue is bent over so far that the underside of the tongue articulates with the palate.

    "Retroflex" in the strictest sense is just 3), which occurs in southern India. But most commonly, it's used as a cover term for all three.

    I am very skeptical one could generate experimental support for the idea that "native MSM speakers […] produce more of a /ʒ/ sound for the initial r of rènshi compared to more of a /r/ sound for the initial r of rén."

    I agree. Instead, the unstressed position between two vowels in the middle of nà liǎng ge rén may cause a preference for realizations closer to the approximant end [ɻ] of the range of r, while the stressed word-initial position in rènshi may cause a preference for the fricative end [ʐ].

    Often the sound ends up somewhere in the middle between these extremes.

  12. Jonathan Smith said,

    April 26, 2019 @ 11:19 pm

    @ David Marjanović
    Re: Mandarin, now actually looking at close descriptions as opposed to spouting off, Lee (2008:109) says that "'sh zh r' are not retroflexes, but apical or upperapical postalveolar, i.e. ʃ̺ tʃ̺ ɹ̺" and Proctor et. al. (2012:214-215) say that Mandarin sibilants generally "are not retroflexed" and refer rather to "bunched coronal" and to laminal coronal gesture (using ʃ̻ for most speakers). So at any rate "retroflex" does not seem to be a favored term in this context, although I am happy to accept your claim that it is technically accurate on some understandings.

    Re: allophones, yes stress and/or intervocalic position could be involved, but I don't have any sense of the postulated difference and so would want to see something concrete. Re: r-, I continue to find representations like [ʐ]/[ʒ] fairly awful… [ɻ] can serve in a pinch.

  13. Chris Button said,

    April 27, 2019 @ 12:20 am

    @ Michael Watts & Jonathan Smith respectively

    The pinyin r is conceptually the voiced equivalent of pinyin sh

    Re: r-, I continue to find representations like [ʐ]/[ʒ] fairly awful… [ɻ] can serve in a pinch

    Here's Pulleyblank's (1984) succinct account:

    "… it has often been identified as a voiced fricative by foreigners since it first appeared in the language in the Tang period and is still treated in that way in some romanization systems such as Wade-Giles, where j was adopted on the analogy of its value [ʒ] in French"

  14. Ignatius Q. Schnockleboster said,

    April 27, 2019 @ 1:19 am

    @Phillip Taylor:
    Indeed, in its original (traditional) form, the "rèn" in "rènshi" (「認」 as in "recognize") has nothing to do with the "rén" (「人」 as in person). Their similarity in pronunciation is not from the similarity of the characters; the opposite is more likely, that the people developing simplified Chinese crafted the simplified 「认」 because of its similarity in pronunciation to 「人」.

  15. Philip Taylor said,

    April 27, 2019 @ 8:24 am

    Ignatius, thank you. That is very interesting — I was previously unaware of just how different the traditional character for "rènshi" is from the simplified version. It would be interesting to compare audio recordings of 「認 that pre-date the introduction of simplified characters with analogous recordings made today to see whether "spelling pronunciation" might affect (some) Chinese speakers in the same way that it affects (some) Western.

    All : the audio segments are now at http://hellenic-institute.uk/audio/Chinese/ in ".wav" format; if anyone cannot handle .wav files, I can easily convert to other formats.

  16. Jonathan Smith said,

    April 27, 2019 @ 3:32 pm

    Ah thanks @ Philip Taylor. Unfortunately here we have different speakers (of different genders) so I'm afraid no conclusions along the lines you suggest can be drawn.

    To me the crucial things about Mand. onset r- are:

    * it's rhotic, a class whose unified nature is difficult to characterize but perhaps concerns in part a combination of tongue root and tip/body gestures (see this thesis, etc.) This is the crucial quality which is lost in representations like ʐ or ʒ. So IMO pinyin "r" is a very good spelling choice, and English "r"(s) are a good basis for approaching the Mandarin sound. BUT

    * the forward/coronal gesture different — not "bunched" like most AmE and certainly not retroflex senso stricto, but akin to the zh/ch/sh gesture for which see discussion above. To me "laminal" seems like a good characterization but YMMV

    * no lip rounding/bunching as in (most?) English onset /r/. Subtle but dead give-away for non-native speech. It is instructive to watch native speakers' lips for syllables like rang, with which contrast English rang/wrong/etc ignoring vowel quality. Of course this changes given Mand. r + u/w. And finally…

    * variable turbulence/frication at the front tongue contact. Your male speaker has more in this recording, the female minimal. This is of course the feature that screams z/ʐ/ʒ to non-natives. A learner can dispense with this entirely… although my saying so might be a response to a slight sociolinguistic cline along which heavy frication is low-end…

  17. Philip Taylor said,

    April 28, 2019 @ 11:59 am

    Jonathan, many thanks for your most helpful and informative analysis. I have added a third recording from the same series, "Dialogue.wav", in which one can hear the same speaker (S1) utter both rènshi (U1) and rén (U3) and compare this with her collocutor (S2) uttering both rènshi (U2) and rén (U4). To my ear, S1's r in rènshi is closer to /r/ whilst S2's is closer to /ʒ/, but S2's r in Yīngguórén is (to my ear) a clear /r/. I would be interested in your further analyses of these, if you have both the interest and the time.

  18. Jonathan Smith said,

    April 29, 2019 @ 7:16 am

    Hi Philip Taylor, I am still inclined to hear the main difference as between S1 with clearer and S2 with more "turbid" realizations of r-… but you could very well be right that S2's two tokens are also different along these lines… if so we would probably want to refer to factors noted above like articulatory effort ~ stress rather than to some difference inherent to the two words in question — but experimentation on the matter could certainly prove me wrong! My thoughts ~~

  19. Philip Taylor said,

    April 30, 2019 @ 7:32 am

    Thank you once again, Jonathan. When I have time I will try to seek out and upload some further examples of apparently contrasting spoken renderings of Pinyin r (ideally by the same speaker)1, but in the meantime I do think that this discussion has tended to support the very point that I sought to make in my opening contribution — that Hanyu Pinyin is a very broad phonemic transcription system (rather than a phonetic one) for spoken MSM , and that when discussing (e.g.,) the various allophones of r, the use of a narrower, more phonetic, transcription system such as the IPA would seem to have considerable potential merit.
    ——–
    1I am thinking of the dialogue in which Rachael says Stuart, ràng wǒ jìeshào yīxià. Here we have phrase-initial r followed by falling-vowel, just as we have in rènshi nĭ, but I still have to locate a passage in which Rachael uses rènshi in a similar position …

  20. Dia said,

    May 2, 2019 @ 4:04 am

    I just discover this blog. As a native Mandarin speaker, I learn the speaking of each word from Phonetic/Bopomo system.
    In my personal opinion, Pinyin system does not accurately translate every spelling of Mandarin into spoken language. For a new learner who speaks English (or recognizes Latin) already, Pinyin sometimes leads to mis-read, somewhat like Jonathan Smith said.

  21. Philip Taylor said,

    May 2, 2019 @ 8:30 am

    Dia, I find your statement very interesting because Wikipedia suggests that the two systems (bopomofo/zhuyin, pinyin) are isomorphic :
    "Zhuyin and pinyin are based on the same Mandarin pronunciations, hence there is a one-to-one correspondence between the two systems". What is it about bopomofo/zhuyin that, in your experience/opinion, enables it to more "accurately translate every spelling of Mandarin into spoken language" ? Is it simply that the Western "natural" pronunciation of the pinyin alphabet is a (very) poor guide to how it should be pronounced in the context of Mandarin, or is it more complex than that ?

  22. Chris Button said,

    May 2, 2019 @ 8:09 pm

    @ Philip Taylor

    Regarding structural differences between pinyin and bopomofo, we touched on that a little here:

    http://languagelog.ldc.upenn.edu/nll/?p=40892#comment-1557742

  23. David Marjanović said,

    May 3, 2019 @ 6:07 am

    Lee (2008:109) says that "'sh zh r' are not retroflexes, but apical or upperapical postalveolar, i.e. ʃ̺ tʃ̺ ɹ̺" and Proctor et. al. (2012:214-215) say that Mandarin sibilants generally "are not retroflexed" and refer rather to "bunched coronal" and to laminal coronal gesture (using ʃ̻ for most speakers).

    Interesting that one says it's apical while the other says it's laminal. I suspect both are true for different speakers and I simply don't know enough about the variation within Mandarin (as usual).

    This distinction of places of articulation on the tongue is orthogonal to that of places of articulation on the palate. However, apical postalveolars (same place on the palate as English /ʃ/) definitely sound more retroflex than laminal ones. Tiny amounts of lip-rounding also have a major impact on how "back" postalveolars sound.

    […] Pinyin system does not accurately translate every spelling of Mandarin into spoken language. For a new learner who speaks English (or recognizes Latin) already, Pinyin sometimes leads to mis-read, somewhat like Jonathan Smith said.

    The second sentence is of course true – learning Pinyin means to learn a new orthography, same as for almost any other language that uses the Latin alphabet; I don't see what impact that has on the first sentence, though, and I disagree with the first in that all distinctions Standard Mandarin makes are represented unambiguously in Pinyin (as also in Bopomofo and most other transcription systems).

  24. IA said,

    May 3, 2019 @ 8:21 am

    YR Chao: Mandarin Primer (1948):

    'Of the other three columns, only the initial r- needs comment. The Chinese r differs from English r in two respects. […] The other difference is that it is shorter and has more friction. That is why the Wade system uses the letter j (as in French je) for this sound. However, since the speakers of many Chinese dialects pronounce this sound with no friction, the English pronunciation will result in no "foreign accent" in this respect. '

  25. IA said,

    May 3, 2019 @ 8:37 am

    In Taiwan, there a few people who carry out some sort of hypercorrection on initial /r/.

    What they do is, for example, say 「日本」such that it sounds like「治本」(but with voiced initial),「仁愛路」such that it sounds like「zhén 愛路」, and so on.

    A noticeable and notable perpetrator is the television news anchor 張雅琴.

    (Or maybe hypercorrection is not an accurate characterisation, I don't know. What would they be correcting? Lack of perceived friction, and as a result turning a fricative into an affricate?)

    Anyway, does this also happen anywhere 'on the mainland'? (Or 'in China' if you prefer.)

  26. Jonathan Smith said,

    May 3, 2019 @ 9:16 am

    For use as an orthography for Mandarin, any approximately phonemic system is fine; as far as specific Roman-letter choices, as I said above, I think pinyin's "r" is the best one for the sound in question. Narrower phonetic transcription/description can of course useful for learners but doesn't belong in regular spellings IMO.
    @IA Looks like I agree with YR Chao here.
    I listened to some of 張雅琴; some realizations of r- indeed sound "affricated". This is totally new to me. No I wouldn't call this hypercorrection. Maybe derhoticization is on the horizon here.

RSS feed for comments on this post