Ted Chiang uninvents Chinese characters

« previous post | next post »

Yesterday morning on the commute to Penn, I was intrigued by a series of six articles in the latest New Yorker (5/16/16) that appeared under the rubric "Uninvent this":  Mary Karr on high heels, Charlie Brooker on dancing, Carrie Brownstein on conference calls, Lee Child on fiction, Alexandra Kleeman on mirrors….  When I reached the sixth and last one, I was so stunned that I almost dropped the magazine and nearly fell out of my seat.

The article on p. 77 that struck me so powerfully was "Bad Character" by Ted Chiang.  The very idea!  To uninvent Chinese characters!

Ted Chiang is a science fiction writer.  Only a science fiction writer or a clairvoyant linguist would have the audacity to imagine a world without Chinese characters.  Yet that is exactly what Ted Chiang does in his brief, but densely packed, essay.

At first I thought I would excerpt portions of the essay, but the whole thing is so brilliant and cogent that it is hard to decide which parts to omit.  Furthermore, the essay is short enough that it will not take up too much space in this post.

It’s not personal. I never learned anything in the Saturday-morning Chinese school I was forced to attend as a child, but that’s not what motivates my choice here. There were plenty of reasons for my poor performance in those classes—my resentment at having to miss the “Super Friends” cartoon being just one of them—so I don’t blame Chinese characters for my failure.

No, my objection is a practical one: I’m a fan of literacy, and Chinese characters have been an obstacle to literacy for millennia. With a phonetic writing system like an alphabet or a syllabary, you need only learn a few dozen symbols and you can read most everything printed in a newspaper. With Chinese characters, you have to learn three thousand. And writing is even more difficult than reading; when you can’t use pronunciation as an aid to spelling, you have to rely on pure memorization. The cognitive demands are so great that even highly educated Chinese speakers regularly forget how to write characters they haven’t used recently.

The huge number of characters poses other obstacles as well. I’ve flipped through a Chinese dictionary, I’ve seen photographs of a Chinese typewriter, I’ve read about Chinese telegraphy, and despite their ingenuity they are all cumbersome inventions, wheelbarrows for the millstone around Chinese culture’s neck. Computers and smartphones are impossible to use if you’re restricted to Chinese characters; it’s only with phonetic systems of writing, like Bopomofo and Pinyin, that text entry becomes practical. In the past century, there have been multiple proposals to replace Chinese characters with an alphabet, all unsuccessful; the only reform ever implemented was to invent simplified versions of the more complex characters, which solved none of the problems I’ve mentioned and created new ones besides.

So let’s imagine a world in which Chinese characters were never invented in the first place. Given such a void, the alphabet might have spread east from India in a way that it couldn’t in our history, but, to keep this from being an Indo-Eurocentric thought experiment, let’s suppose that the ancient Chinese invented their own phonetic system of writing, something like the modern Bopomofo, some thirty-two hundred years ago. What might the consequences be? Increased literacy is the most obvious one, and easier adoption of modern technologies is another. But allow me to speculate about one other possible effect.

One of the virtues claimed for Chinese characters is that they make it easy to read works written thousands of years ago. The ease of reading classical Chinese has been significantly overstated, but, to the extent that ancient texts remain understandable, I suspect it’s due to the fact that Chinese characters aren’t phonetic. Pronunciation changes over the centuries, and when you write with an alphabet spellings eventually adapt to follow suit. (Consider the differences between “Beowulf,” “The Canterbury Tales,” and “Hamlet.”) Classical Chinese remains readable precisely because the characters are immune to the vagaries of sound. So if ancient Chinese manuscripts had been written with phonetic symbols, they’d become harder to decipher over time.

Chinese culture is notorious for the value it places on tradition. It would be reductive to claim that this is entirely a result of the readability of classical Chinese, but I think it’s reasonable to propose that there is some influence. Imagine a world in which written English had changed so little that works of “Beowulf” ’s era remained continuously readable for the past twelve hundred years. I could easily believe that, in such a world, contemporary English culture would retain more Anglo-Saxon values than it does now. So it seems plausible that in this counterfactual history I’m positing, a world in which the intelligibility of Chinese texts erodes under the currents of phonological change, Chinese culture might not be so rooted in the past. Perhaps China would have evolved more throughout the millennia and exhibited less resistance to new ideas. Perhaps it would have been better equipped to deal with modernity in ways completely unrelated to an improved ability to use telegraphy or computers.

I have no idea if I would personally be better off in such a world, assuming that it’s even meaningful to talk about my existing there at all. But there is one thing I’m certain of: in a world where Chinese was written with phonetic symbols, I would never have to read or hear any more popular misconceptions about Chinese characters—that they’re like little pictures, that they represent ideas directly, that the Chinese word for “crisis” is “danger” plus “opportunity.”* That, at least, would be a relief.

* "danger + opportunity ≠ crisis:  How a misunderstanding about Chinese characters has led many astray"

When I read the penultimate sentence of Chiang's essay, I experienced an epiphany:  I knew for certain that I had encountered a true zhīyīn 知音 (lit., "knows / understands / appreciates sound", i.e., someone who understands what I've been saying [for the past half century]).

The whole thrust of Chiang's argument is on the possibilities that are inherent in the sounds of Chinese, as opposed to the difficult, demanding, nonphonetic characters.  It is fitting that the link for this article, which I have embedded in the title, has as its tag identifier "if-chinese-were-phonetic".

Vox clamantis in deserto!

Kōnggǔ zúyīn 空谷足音 ("the sound of footsteps in a deserted valley")!


  1. Bîp said,

    May 13, 2016 @ 2:13 pm

    Stibna wopjandins in auþidai!

  2. Coby Lubliner said,

    May 13, 2016 @ 2:33 pm

    Another possibility is that there might be no "China": the various topolects would, like Romance, have developed into independent "languages" and formed distinct national identities based on them.

  3. liuyao said,

    May 13, 2016 @ 4:30 pm

    Thought-provoking. If Chinese had adopted phonetic writing early on, very little of anything would be left that is recognizably Chinese (and I'd like to hear about what the Chinese languages would have become). Given the language diversity, a unified China may not last longer than the Roman Empire, and one could speculate that the Chinese may actually have discovered America first, colonized Europe, and we'd all be typing in the Chinese alphabet today (perhaps a particular language from a coastal region or offshore isles) in discussing all matters. Even if the Chinese did not discover deductive reasoning of the Greeks, the prerequisite to modern scientific revolution and the subsequent industrial revolutions, they'd surely be able to pick it up much more easily. That is, all the things that characters were blamed for in late-19th / early-20th century intellectual debate in China.

    I'd agree with much of the outcome of the thought experiment. I'm hesitant though to put judgement values to it, that it'd be in many respects "much better." The argument about literacy tends to oversimplify the matter, making it sound as if learning the 26 symbols and how they combine to make sounds would automatically give you the ability to read newspapers, and it's a contrast of 26 vs 3000, or more. If that were the case, why is it that Chinese language courses don't take 100 times longer than other languages? Why didn't the Chinese simply abandon characters at the first encounters with Buddhist scriptures, or later the scripts of the Mongols and Manchus? If an objective difficulty index of language+script combination is devised, Chinese (any of its topolects) with characters shouldn't be more than twice as difficult than a purely phonetic language, no?

    [A question: if we are dealing with the whole span of human history, was literacy much higher in medieval Europe? Socio-economic factors are perhaps more important, as amply debated here.]

    Also, the majority of Chinese people these days know little, and have abandoned much of their tradition, and can't genuinely have much respect for it even if they claim so; only the characters live on. Perhaps that's precisely why they are so attached to characters, without which they'd have nothing left to claim to be Chinese.

  4. Chris C. said,

    May 13, 2016 @ 6:53 pm

    "to keep this from being an Indo-Eurocentric thought experiment"

    One of the few things to criticize in this essay. Chaing seems unaware that the alphabet is a Semitic invention. (Or, if we want to pick nits, the alphabet is but a minor adjustment of what is fundamentally a Semitic idea.)

    I'm also not so sure how accurate is his characterization of China being resistant to change. You only have to see the list of common modern technologies that were invented in China to appreciate that for a long time it was a highly inventive culture, from a time when the writing system was already well-entrenched. Cultures turn inward for any number of reasons besides conservatism in the form of the written word. An ideology where China is the "Middle Kingdom", the center of the world, self-sufficient and needing nothing from anyone outside, has at least as much to it as the writing system, I'd think. And that's without even taking into account the shock of direct contact with the West, as opposed to contact through the intermediary cultures of Central Asia as had been the case for centuries.

    But I'm talking out my ass at least a little bit here, so I'll stop.

    If Chinese had been written phonetically, I wonder how many ancient puzzles would be less puzzling? I have in mind those which partly have to do with transcriptions of foreign names into a Chinese which must itself be reconstructed.

  5. Jim Breen said,

    May 13, 2016 @ 6:55 pm

    Thanks for posting the article, Victor. I enjoyed it considerably.

    Amen to your dig with 知音. I often quote that word when advising people NEVER to attempt to deduce the meaning of a compound from the constituent kanji/hanzi. Another one, in Japanese at least, is 魂柱.

  6. Victor Mair said,

    May 13, 2016 @ 8:13 pm


    I did not know this word; had to look it up.

    The characters literally mean "soul column / post / pillar". In Japanese they would be pronounced konchū or tamabashira. The actual meaning of the word is "sound post" of a stringed instrument.

    From Wikipedia:


    In a string instrument, the sound post or soundpost is a small dowel inside the instrument under the treble end of the bridge, spanning the space between the top and back plates and held in place by friction. It serves as a structural support for an archtop instrument, transfers sound from the top plate to the back plate and alters the tone of the instrument by changing the vibrational modes of the plates.

    The sound post is sometimes referred to as the âme, a French word meaning "soul". The bow has also been referred to as the soul of these instruments. The Italians use the same term, anima, for this.



  7. Mara K said,

    May 13, 2016 @ 11:06 pm

    First thought: I disagree that a phonetic system would unmoor China from its literary heritage. As I argued to someone on Facebook with whom I shared this post, I think a phonetic writing system for Chinese would make the classics even more accessible to the masses.

    Second thought: What does Chiang mean by "Anglo-Saxon values"? Given that he's the author of one of my favorite science fiction short stories, I want to see him do an alternate history in which English uses characters and Mandarin has an alphabet.

  8. Bob Ladd said,

    May 14, 2016 @ 12:24 am

    @Chris C.: I read "Indo-Eurocentric" not as referring to Indo-European (which you apparently did), but as referring to the whole civilisation whose roots lie in the area from India to Europe – which would include all those ancient Semites who helped develop writing. Either way, though, if we're quibbling about credit for inventing the alphabet, then (1) it was arguably the Sumerians who first got the basic symbols-for-sounds idea, and the Sumerians weren't Semitic; (2) the innovation of representing consonants and vowels with separate symbols was an Indo-European (specifically Greek) idea. To me, breaking free of symbols-for-syllables or symbols-for-CV-sequences was just as important a step as getting rid of symbols-for-morphemes, and it was that second step that gave us the alphabet.

  9. John Cowan said,

    May 14, 2016 @ 1:40 am

    English uses characters


  10. Sean M said,

    May 14, 2016 @ 4:39 am

    @Bob Ladd: Just keep in mind that cuneiform had signs for /a/ /e/ /i/ /u/ from the third millennium BCE onwards and used them for things like "the Akkadian conjunction /u/" and "that last sign could be pronounced /he/ or /hi/ so I'll write an /e/ to make it clear which I mean." Aramaic uses one sign for Halbvokalen and their corresponding vowel, but so does the ancient Latin alphabet, and when signs can be read as a consonant or a vowel, the vocalic reading is Greek is usually one which the corresponding sign in Aramaic has https://bookandsword.com/2015/03/28/from-aleph-bet-to-alphabet/ How the Northwest Semitic abjads spread to the Aegean is not an area of my expertise, but pinning down what the Greek alphabet does differently is tricky … they were sure fussier about distinguishing vowels than any earlier script!

  11. Chris Button said,

    May 14, 2016 @ 5:53 am

    Very interesting!

    As I commented in an older thread, readers of English and readers of Chinese have far more in common than readers of English and for example readers of Spanish. That is of course not to suggest that learning to read/write Chinese and learning to read/write English are of a comparative level of difficulty.

    Regarding the emergence of an alphabet/syllabary, it might be worth noting Victor's and Edwin Pulleyblank's comments regarding a possible early role for the Tiangan Dizhi that was not pursued in China's orthographic tradition.

  12. David Marjanović said,

    May 14, 2016 @ 6:18 am

    Another possibility is that there might be no "China": the various topolects would, like Romance, have developed into independent "languages" and formed distinct national identities based on them.

    The reason for that, I think, is that the kingdoms established by the various barbarians remained stable. This led to the emergence of new identities: France is defined by the king of the Franks, Spain is defined by the king of the Visigoths. That never happened in China.

    It took easily 500 years after the collapse of the Western Roman Empire till Romance-speaking people stopped thinking of their languages as simply Latin.

    Why didn't the Chinese simply abandon characters at the first encounters with Buddhist scriptures, or later the scripts of the Mongols and Manchus?

    Because of the weight of tradition in the literate = ruling class. See also: Babylonians and Elamites not abandoning cuneiform (with signs for words and syllables, much like Japanese) after contact with Aramaic and Old Persian, Cypriot Greeks holding on to Linear C (signs for vowels and CV syllables) for several hundred years after introduction of the Greek alphabet.

    was literacy much higher in medieval Europe?

    In some places. In Scandinavia and the Republic of Novgorod, it does look like every idiot could read and write (much like in ancient Pompeii, BTW). Elsewhere, literacy was so strongly linked to Latin and the church that it didn't spread far.

  13. chris said,

    May 14, 2016 @ 7:37 am

    to keep this from being an Indo-Eurocentric thought experiment, let’s suppose that the ancient Chinese invented their own phonetic system of writing, something like the modern Bopomofo, some thirty-two hundred years ago

    Wouldn't it be simpler to assume that China adopted one of the Japanese systems rather than vice versa? (Of course if Chinese characters don't exist, the Japanese can't mix them into their own writing system.)

    @Cory Lubliner: Seems like a reasonable speculation, but on the other hand, Sanskrit *did* split up into separate languages and India is an identifiable country today. A common language isn't indispensable.

  14. Tom said,

    May 14, 2016 @ 8:05 am

    And there's this response to Ted Chiang from Tom Mullaney: http://foreignpolicy.com/2016/05/12/chinese-is-a-21st-century-language-ignore-orientalism-2-0-critiques/

  15. Mike Williams said,

    May 14, 2016 @ 8:25 am

    Ted Chiang is a fabulous writer of short fiction. Most of his stories are available online. If you're interested to read more, they can be found via https://en.wikipedia.org/wiki/Ted_Chiang#Works. I loved "The Merchant and the Alchemist's Gate", but readers of Language Log might particularly enjoy "Story of Your Life".

  16. Jonathan Badger said,

    May 14, 2016 @ 9:08 am

    @Mara K:
    Maybe cremation would have been historically more popular in English speaking countries thanks to Beowulf's choice as to the disposal of his remains. And we would be meeting our coworkers for a drink after work in our local mead hall rather than a bar. But seriously, I think the language barrier makes English speakers see the Anglo-Saxons as a fundamentally different culture from ourselves in a way that we don't see with say, Shakespeare.

  17. Victor Mair said,

    May 14, 2016 @ 9:20 am

    Did Tom Mullaney call Ted Chiang an Orientalist?

  18. Bob Ladd said,

    May 14, 2016 @ 10:59 am

    @Victor Mair (Did Tom Mullaney call Ted Chiang an Orientalist?)
    In effect, I think he did. He certainly implies that he doesn't know how Chinese characters really work, which anyone can see from the New Yorker essay is simply false.

  19. Coby Lubliner said,

    May 14, 2016 @ 11:02 am

    I agree with David Marjanović that what I said about languages and nations was an oversimplification. Still, China was also divided at various times into several states, but the sense of Chinese identity never went away, and that was probably what made it possible for the Mongol Kublai Khan, an admirer of Chinese culture, to reunify the empire. By contrast, what was left of the shared Latin culture of Europe was not enough to make something like that possible there. Even Napoleon, who would have liked to rule over all Europe, had to create nominally independent kingdoms for his brothers and friends.

  20. Coby Lubliner said,

    May 14, 2016 @ 11:05 am

    @ chris: I think Hinduism is the unifying force in India.

  21. Thorin said,

    May 14, 2016 @ 2:17 pm

    @Mike Williams I'm waiting for the film version of Story of Your Life to be released this year. Very excited!

  22. JK said,

    May 14, 2016 @ 2:29 pm

    "Computers and smartphones are impossible to use if you’re restricted to Chinese characters; it’s only with phonetic systems of writing, like Bopomofo and Pinyin, that text entry becomes practical."

    Isn't this an incorrect statement? I always thought the fastest input methods were the stroke-based ones, and now most smartphones have character recognition for handwriting.

  23. Victor Mair said,

    May 14, 2016 @ 2:37 pm


    Such CLAIMS have been made for stroke-based inputting for decades, but still the vast majority of users stick with phonetic inputting. There must be a reason why they do so.

    We have been over this time and time again on Language Log.

  24. Chris C. said,

    May 14, 2016 @ 2:44 pm

    @chris — All the Japanese writing systems are based on the Chinese, including kanji and both types of kana. China is where Japan learned writing, possibly by way of Korea, so whatever writing system China had settled on, Japan would follow.

  25. Chris C. said,

    May 14, 2016 @ 2:59 pm

    @Bob — "it was arguably the Sumerians who first got the basic symbols-for-sounds idea" — Cuneiform was a mixed system, much as Chinese is today, and as purely phonetic symbols it was a syllabary (usually), not an alphabet. It was the observation that, unlike systems of writing devised expressly for Semitic languages, cuneiform definitely represented vowels, which provided one of the most important clues that an older, then-unknown language lay behind it, when most of the cuneiform corpus of the day (mid-19th century) was Akkadian.

    "the innovation of representing consonants and vowels with separate symbols was an Indo-European " — I'm glad you picked up on what I took the trouble to imply.

  26. JK said,

    May 14, 2016 @ 3:03 pm

    @Victor Mair

    His statement is still incorrect, at least the first part: "Computers and smartphones are impossible to use if you’re restricted to Chinese characters." As for the second part, I don't know of any Chinese text input method that is solely phonetic. They are partially phonetic, but you still have to choose a character from a list that pops up, and trying to put in tones would probably slow people down. Most of the time you can just type part of the pinyin and directly go to choosing the character, so I could easily argue that the pronunciation input is helpful, but the critical part of getting the right character to show up on my screen is picking out the one I want.

    I think this is where the critique of Orientalism comes in (not that I am making that argument myself) — "it’s only with phonetic systems of writing, like Bopomofo and Pinyin, that text entry becomes practical" still sounds ridiculous to me. Why not say that it is only with computer databases that intelligently associate the characters I want to type with partial pronunciation input that text entry becomes practical?

  27. Victor Mair said,

    May 14, 2016 @ 3:25 pm


    I type whole paragraphs with phonetic inputting. Seldom do I ever have to look up a single character.

  28. JK said,

    May 14, 2016 @ 3:30 pm

    @Victor Mair

    I'm sorry, I don't follow. Are you typing paragraphs to be displayed in pinyin, or do you mean the computer association of characters with your pinyin input is so good that you never have to correct it? If the latter is the case, my whole point is that the computer association of the characters to your pinyin input is the critical part, not that your input is in pinyin.

  29. Victor Mair said,

    May 14, 2016 @ 3:36 pm


    Yes, the computer association of the characters to the input is that good, so good that I am constantly stunned by how the computer picks the right characters most of the time.

    As do most people, I choose to input with pinyin because it is so much easier and faster than stroke order inputting.

    In most cases, my desired output is Chinese characters, so naturally that's what I want the pinyin to do for me via the computer.

  30. Ross Bender said,

    May 15, 2016 @ 1:01 pm

    When I read the New Yorker article all of two days ago I said to myself, "Boy, Victor Mair is gonna love this one, lah!" And sure enough, here we are.

    It is probably already too late to make a comment in this highly feverish modern medium — the weblog, if they still call it that (just for reference today is May 15, 2016 in the Gregorian Calendar), but I just wanted to recommend highly Tom Mullaney's article cited above — "Chinese is Not a Backward Language."

    In my college years I read Hegel's Philosophy of History in English translation and discovered that the great philosopher had decreed that Chinese culture was "the infancy of the race," which was clearly evidenced by the fact that the Chinese language sounded like baby talk. Mullaney quotes Hegel further on the subject, in his discussion of Orientalism 1.0.

    Hegel pointed out that of course Africa had no history, then went on to say that India was the "dreamy adolescence" of the Race, Greece and Rome the "sturdy youth", and of course that (you guessed it) modern Germany was the maturation of the Race, the pinnacle of Civilization (in particular, I believe he specified Prussia).

    There is much more in Mullaney's excellent article and I would hope that all LanguageLog readers would find time to read it. Mullaney cites Herrlee Creel's 1936 essay "On the Nature of Chinese Ideography" as a corrective to Orientalism 1.0, then goes on to mention the work of Geoffrey Sampson and Jack Goody in this context.

    There is a tendency among some in the ongoing orality/literacy discourse to presume that the phonetic alphabet is the supreme orthographic pinnacle to which all civilizations aspire. Part of that line of argument is that the Greeks, upon inventing the alphabet, stimulated a massive change in brain function that led, among other things, to TECHNOLOGY, and of course our brilliant Western Civilization, the acme, zenith, epitome, and ne plus ultra of all the best in the whole damn universe.

    Having studied classical Chinese with Victor Mair and having attended his lectures on Chinese script and society, I am well aware of his personal enthusiasm for pinyin. While he is the last person I would accuse of Nasty Orientalism, I would be very interested in his response to Mullaney's essay.

    Creel's article, btw, was in T'oung Pao, Volume 32, Issue 1, pages 85 – 161 Publication Year : 1936 DOI: 10.1163/156853236X00056.

    I could of course trudge up to the library and attempt to find it in the musty old stacks, but in this ultramodern age I simply went to the library website and ordered a digital copy, which I expect will arrive momentarily.

    While Creel's (and Sampson's and Goody's) views are old hat (good grief, this IS 2016, after all, and after all we're practically in the fourth millennium AD or CE or the Age of Aquarius or something, I have much respect for all three scholars and would very much appreciate a slow read and exposition of their ideas.

    And now, as I am old and it is Sunday afternoon in Philadelphia, it is time for my nap.

  31. Chris Button said,

    May 16, 2016 @ 12:31 pm

    Somewhat conversely to Ted Chiang's imaginary world without Chinese characters, I have sometimes wondered about a world where the "alphabetic" consonant-vowel dichotomy had never emerged as anything outside of purely scientific labels applied by phoneticians to particular acoustic phenomena.

    People who have made the argument that vowels don't really exist beyond a schwa and a stressed variant have often hit an academic wall in linguistics. No matter how well argued the case may be, the very idea of such a notion in an academic discipline thoroughly dominated by "Western" notions of a consonant-vowel dichotomy is almost sacrilegious. As a result, a plethora of vowels in neat triangular systems (that undoubtedly exist at the surface level and can be measured by phoneticians) are still regularly crammed into the underlying structure of reconstructed languages like Indo-European and Sino-Tibeto-Burman when the evidence quite clearly points in the other direction. In the specific case of Old Chinese here, the relationships between xiesheng series of characters (i.e. those written with the same "phonetic") only become clear once a basic, stress-conditioned, vertical schwa/"a" ablaut is established. It is this schwa/a ablaut that makes strings of consonants pronounceable in what essentially constitutes the formation of the "syllable" rather than amounting to any notions of a "vowel system". This "syllable" is the most basic concept of language that is comprehensible to any speaker whether they are literate or not.

    While I am sympathetic to the relative ease of using an alphabet in promoting literacy (and am not necessarily arguing here against its promotion), it is nonetheless an artificial abstraction that separates the user's inherent sense of syllable structure from the written language. As a result, it is English spelling with all its irregularities (conditioned by enormous foreign influence and by the ever-shifting spoken word, combined with a reluctance to modify spelling as appropriate to maintain a degree of "regularity") that is driving the reader closer to a syllabic usage of English writing. In essence, the English alphabet encourages people to read it in a similar manner to how they would read the Chinese script. In that sense, any notion of alphabetic superiority goes out the window as English writing has essentially already reverted back to a more "natural" Chinese system.

  32. JS said,

    May 16, 2016 @ 1:57 pm

    @Chris Button
    Interesting comment! I appreciate your point that the syllable, as opposed to the segment (whether "consonant" or "vowel"), is a or the psychologically fundamental phonological unit. Thus your claim is not without basis that the "consonant-vowel dichotomy" imposed by an alphabet is a contrived one.

    But it doesn't at all follow that reconstructions with minimal vowel systems are to be preferred, as if the history of language in general over the past several thousand (?) years might have involved steady diversification away from more limited and "natural" one- or two-vowel arrangements. On the contrary, this position reflects a very odd theoretical commitment regarding the very segments whose primacy you are arguing against. To put it another way: traditional "reconstructions" of OC are of course in terms of segment-less rhyme categories, and many Chinese historical phonologists continue to work in such terms; given your stated skepticism towards segmental analysis, why not do the same? To me, describing ~30 such categories with only two vowels is a marginally interesting academic exercise that points to exactly the opposite mindset.

  33. Victor Mair said,

    May 16, 2016 @ 2:10 pm

    It's sad that, after taking my semester-long course on "Language, Script, and Society in China", taking a rigorous year-long introduction to Literary Sinitic with me, and reading the kanbun of the Shoku Nihongi together for a couple of years, someone would still think that Orientalist discourse has something to do with Chinese linguistics.

    On the other hand, knowing the author and judging from the last, ironic paragraph, it's quite possible that his entire long comment was written tongue in cheek.

    Here is a good follow-up to this post about Ted Chiang's uninvention of Chinese characters and the ensuing discussion:

    "Backward Thinking about Orientalism and Chinese Characters" (5/16/16)


  34. Chris (different Chris) said,

    May 16, 2016 @ 2:31 pm

    Hm. So is Iceland conservative and resistant to change because modern Icelanders can (if I understand correctly) still read their foundational literature without translation?

    Along the same lines, does Iceland retain ancient social values? Do modern Icelanders consider themselves part of the same culture as their distant ancestors?

    I'm not trying to be snide, I'm skeptical but asking sincerely.

    (Admittedly, the values of "foundational" and "distant" are substantially different from what they might be in China by at least a couple of thousand years….)

  35. Chris Button said,

    May 16, 2016 @ 2:58 pm


    Thanks for your comment. The suggestion is not that the evolution of language has "involved steady diversification away from more limited and "natural" one- or two-vowel arrangements." The suggestion is rather that underlyingly the notion of a "vowel" does not exist in spite of the multiple vowels that may be heard on the surface at any stage in the history of language.

    A "vowelless" analysis can be made for modern languages too (and has been attempted), although certain languages just happen to be more amenable to such an analysis from a purely internal perspective than others. Focusing solely on the languages in my STEDT monograph: Northern Chin languages do not tend to support a vowelless analysis; Burmese, when analyzed on the basis of its inscriptional orthography (~1000yrs ago) does support such an analysis; Old Chinese supports such an analysis but only when going back ~3000yrs – i.e. well beyond the time depth of Old Burmese. The time depth itself is immaterial; the point is that when one starts reconstructing a proto-language based on comparative evidence, then "vowellessness" becomes an undeniable component. Northern Chin languages may have "vowelless" origins in Proto-Tibeto-Burman, but they are not vowelless languages (nor on the surface is, or ever has been, any language for that matter).

  36. JS said,

    May 16, 2016 @ 4:58 pm

    Thanks for the clarification. I encourage you to follow your muses, naturally. My only comments would be, one, if "vowellessness" is largely a product of some reconstructive protocol rather than offering new and powerful analyses of living languages, then there's a problem — unless, as was my impression, you believe there was something fundamentally different about ancient languages vis-a-vis modern ones. And two, my point above: if you believe in consonants and in syllables, as seems to be the case, then having vowels or not is just a matter of pushing variables around. Your point about the fundamental nature of the syllable is precisely the point that, to a certain degree, how these variables are arranged is neither here nor there with respect to the theoretical validity of the analysis.

  37. Ross Bender said,

    May 16, 2016 @ 5:17 pm


    Not entirely tongue in cheek. I read the Mullaney essay and thought he might have a point, particularly since he cited Hegel's Orientalist views on Chinese language and culture. While we may dismiss such ideas as entirely antiquated, I do believe it is important to understand how formative and foundational they have been for modern Western thought. After all, Karl Marx inherited Hegelian positions and substituted a materialist dialectic for the former's idealistic dialectic.

    I should say that I am chagrined by my revered teacher's disappointment in my intellectual development. In many ways I consider these Language Log posts to be an online continuation of Victor's very stimulating seminar on "Language, Script, and Society in China", in which we had many vigorous discussions and arguments, and in which I first read the musty old work of Bernhard Karlgren.

    Moser's refutation of Mullaney is certainly devastating. It would be interesting to hear Mullaney's response.

    Just this afternoon I read the Herrlee Creel article "On the Nature of Chinese Ideography." While it would be beyond me to provide an adequate summary of this very learned piece, I was interested to discover that the gist was Creel's critique of Karlgren's emphasis on the phonetic nature of the Chinese script. Creel quotes Karlgren at the outset:

    "As Bernhard Karlgren says: "Away with the old ideography; replace it with phonetic writing."

    Then he comments:

    "One of the chief misconceptions which prevail concerning Chinese is the idea that, if they were only purged of their antiquarian love for the characters, the Chinese could write their language with a phonetic script, perhaps with the Roman alphabet. Even a scholar so thoroughly grounded in Chinese as is Karlgren has made this statement. But he at least is aware that this would be to recast Chinese literary expression entire, practically to make another language; less discerning critics are blithely ignorant of the fact."

    Although Creel covers an immense amount of ground, the crux is his argument against Karlgren's position that "Finally one arrived at a kind of script that was phonetic without being alphabetic" 1). .. . nine tenths of all Chinese characters consist of one 'signific' and one 'phonetic"' 2)."
    1) Karlgren, Philology and Ancient China, p. 33.
    2) Karlgren, Analytic Dictionary, Introduction, p. 4.

    Again, the article was published in 1936 and may seem ridiculously out-dated. Nevertheless it is helpful for me, and I hope it would be for others, to go over the background of the current debates and realize that they have a long and distinguished history.

    Finally, my references to my senescence and frequent naps is simply to voice my disquietude at the dismaying speed of contemporary computerized discourse. I am pleased to see how this thread has continued to develop, particularly with young Button's contributions. The work of historical linguists fascinate me, although I usually find it difficult to keep up with the arguments.

  38. Chris Button said,

    May 16, 2016 @ 9:22 pm

    @ Ross Bender

    Thank you for the kind words. By the way, are you aware of the exchange with Peter Boodberg that Creel's 1936 article engendered: Creel (1938 – TP#34); Boodberg (1937 – HJAS#2, 1940 – TP#35).

    @ JS

    It's not that "there was something fundamentally different about ancient languages vis-a-vis modern ones" but rather that when one analyzes languages as part of a language family as opposed to in isolation then the comparative (and by extension historical) evidence favors the syllable rather than a consonant-vowel dichotomy.

    The suggestion that vowellessness simply involves "pushing variables" (features) around has often been levelled against its proponents. However, when conducted responsibly (i.e. without prior agenda), it is only proposed when the evidence calls for it. Anyone can move around distinctive features and one could easily argue that it is the people force-fitting multiple vowel systems on to Indo-European or Sino-Tibetan who are really guilty of such activity. In the particular case of Old Chinese, the multiple vowels proposed in certain reconstructions often leaves xiesheng (phonetic) series without rhyme nor reason (pun intended).

  39. Tom Vinson said,

    May 17, 2016 @ 8:57 pm

    It is possible for a culture using an alphabetic writing system to maintain considerable continuity over centuries. Most modern Icelanders would have little trouble reading the historical writings of Ari Frodi (Ari the Wise, https://en.wikipedia.org/wiki/Ari_%C3%9Eorgilsson), 11th-12th centuries.
    Granted, most earlier "texts" are poetry, written down centuries later, and often rather cryptic (somewhere between Ulysses and Finnegan's Wake).
    Present-day Icelanders read the old texts using their current pronunciation, which is quite different from that reconstructed for the 9th through 13th centuries. Those of us who speak languages that are recorded in alphabetic scripts often like to think of our notation systems as "phonetic". That's not quite the case.
    Granum salis: the present-day population of Iceland is about 330000, which is a drop in the bucket compared to China.

  40. Ross Bender said,

    May 18, 2016 @ 10:36 am


    Many thanks for the links to the Creel-Boodberg conversation. It is amusing and instructive to read over this debate and particularly to savor the elegant ways in which these giants of the discipline insulted each other. Just a few examples:

    1) Boodberg responding to Creel (HJAS 1937 see full citations of articles below):

    "The extremely valuable suggestions found in Karlgren's numerous works seem to have passed unnoticed by the 'epigraphists.' In the rare cases when these suggestions have attracted attention, their import has been misunderstood and the practical hints contained in them misapplied.2

    2 Cf. particularly Dr. Herrlee Glessner Creel's essay "On the Nature of Chinese Ideography," TP 1936, 85-161, a well expressed, but most ineffectual attempt to demonstrate the unique 'ideographic' characteristics of Chinese script and to combat 'phonological' investigations of archaic graphs. Professor Pelliot's remarks appended to the article rightly condemn Dr. Creel's habit of divorcing writing from the living language. Apart from the author's impossible thesis, one must deplore the general tendency manifest throughout his article (and, alas, too prominently figuring in Sinological research on this continent) of insisting that the Chinese in the development of their writing, as in the evolution of many other of their cultural complexes, followed some mysterious esoteric principles that set them apart from the rest of the human race."

    2) Creel responding to Boodberg (T'oung Pao 1939):

    "The chief reason for Professor Boodberg's confusion is perhaps that same lack of logic which at various points mars his other- wise impressive presentation. Nowhere is this clearer than where he seeks to eliminate from all future discussion of Chinese the term 'ideograph', declaring that "The sooner it is abandoned, the better…." "It the light of his point of view it is interesting that Professor Boodberg's article bristles to an unusual extent with symbols not easily rendered as "words". Let us take a simple case, occurring on his page 333, as follows: G< S1 P1 SP<G1 G <S2 p2 SP< G2 [NB I cannot adequately render Prof. Boodberg's equation in this medium, but it's a doozy – Bender]

    It would be interesting to have Professor Boodberg tell us the phonetic value of the v-shaped line which joins these symbols. It must represent a "word", for he tells us explicitly that "Signs used in writing, however ambiguous, stylized, or symbolic, represent words" 1). Then what word? It cannot, we are told, represent an idea, or even one of several optional words, for a sign cannot represent a word until it is "supplied with the appropriate phoneme"' 2). Even "If we associate with a graph several related words, unable to determine which of them it is supposed to represent exactly, this does not mean that the graph represents the 'idea' or 'concept' behind those words" 3). This leaves us in a dilemma, since "Signs used in writing, however ambiguous, stylized, or symbolic, represent words" 4). If Professor Boodberg was not writing, what was he doing?"

    3) Boodberg responding to Creel (T'oung Pao 1940):

    "With martial stalk, the ghost of 'Ideography' haunts again the platform of sinological Elsinore. In TP 34. 265-294, Professor Herrlee Glessner Creel, taking exception to some statements in the present writer's 'Proleptical Remarks' (HJAS 2. 329-372), directed against his efforts to revive 'ideography' as a method of inter- preting Chinese characters, attempts to re-state his position in a curiously constructed article entitled 'On the Ideographic Element in Ancient Chinese'. Though written in a lively, controversial vein, the paper fails to produce a single new fact which would help us to clarify the important problenm of the structure and genesis of Chinese writing, and enlightens us very little on the nature of 'ideographic' ectoplasm. The article inakes it painfully obvious, on the other hand, that the discussion of the cardinal points of the origin and develop- ment of the Chinese written language cannot be fruitfully continued, so long as the champions of 'ideography' refuse to define what they mean by an 'ideogram', and to explain coherently their reasons for disregarding in specific cases long-established evidence pointing to the phonetic and logooraphic nature of what they term 'ideograms'. One of the chief purposes of my article was not, as Professor Creel imagines, to combat 'ideography' (for I am quite open-miinded on the subject, especially as I do not know what on earth 'ideography' signifies concretely), but to protest against the loose use of the vague term 'ideogram' whenever 'logogram' is meant. It is true, at the same time, that, as a philologist and teacher of Chinese, I am naturally perturbed by-and cannot remain indifferent to- the rise of a methodolooy which produces, not in comparatively innocuous special articles, but in text-books through which a new generation of sinologists is expected to be trained, puerilities such as the following:…"

    Good times, good times!

    Some Proleptical Remarks on The Evolution of Archaic Chinese
    Author(s): Peter A. Boodberg
    Source: Harvard Journal of Asiatic Studies, Vol. 2, No. 3/4 (Dec., 1937), pp. 329-372
    Published by: Harvard-Yenching Institute
    Stable URL: http://www.jstor.org/stable/2717943

    On the Ideographic Element in Ancient Chinese
    Author(s): Herrlee Glessner Creel
    Source: T'oung Pao, Second Series, Vol. 34, Livr. 4 (1939), pp. 265-294
    Published by: Brill
    Stable URL: http://www.jstor.org/stable/4527166

    'Ideography' or Iconolatry?
    Author(s): Peter A. Boodberg
    Source: T'oung Pao, Second Series, Vol. 35, Livr. 4 (1940), pp. 266-288
    Published by: Brill
    Stable URL: http://www.jstor.org/stable/4527182

  41. Chris Button said,

    May 18, 2016 @ 1:11 pm

    @ JS

    I don’t think I fully addressed your following comment: “Your point about the fundamental nature of the syllable is precisely the point that, to a certain degree, how these variables are arranged is neither here nor there with respect to the theoretical validity of the analysis".

    In my studies of Northern Chin languages (NW Burma), there were two things that really jumped out. Firstly their complex verbal inflections seem to derive from the same “-s” suffix that gave the Old Chinese qu-sheng (tone 4 in Mandarin) leaving pairs in modern Mandarin like “mai” (tone 3) ‘buy’ and “mai” (tone4) ’sell’. Secondly, there is a superficial vowel length distinction in syllables usually transcribed with short or long vowels (e.g. lim versus liim). However, as others before me have pointed out, with resonant codas it is not really a vowel length distinction but different weight/stress placement on the vowel or the coda (e.g. limm versus liim). It should be noted that such production occurs in isolation so has nothing to do with prosodic effects resulting from isochronic vowel lengthening. In the case of the resonant codas -j and -w, we are afforded the transcriptional flexibility of -i and -u as some people have used (e.g laj as opposed to lai). While people have tended to differentiate syllables according to vowel length (e.g. laj vs laaj or lai versus laai), one could just as well write laj vs. lai to distinguish the two syllables by marking the distinction on the coda instead of the nucleus. The only other resonant coda that shares this transcriptional flexibility is -w which can be written as -u. For all other resonant codas (-l, -n, -m, -ng, -r) we could perhaps use a diacritic to note a difference but it becomes far more cumbersome.

    In Proto-Indo-European, “j” and “w” are treated like the other resonants. However for the purposes of making an acceptable vowel triangle they are often deceptively treated as “i” and “u” when occurring as the syllabic nucleus (this then nicely parallels the two “e” and “o” vowels of Indo-European while disregarding the fact that, as others have suggested, “e” and “o” are actually schwa and its ablaut variant “a"). In the primordial stages of Indo-European and Old Chinese, a syllable like “pj” or “pn” is perfectly allowable. The schwa (or one could say “syllable") is inherent in the “j” and in the “n". On the other hand, a syllable like “pt” is not allowed until a schwa is inserted to make it pronounceable as “pət".

    The fundamental difference between Old Chinese and Indo-European occurs with something like “pjt” and “pnt". Both are permissible in Indo-European (one might write the former as “pit"), but only the former is permissible in Old Chinese. The reason is because while “j” is a resonant like “n” (as the Northern Chin evidence shows us even in a Sino-Tibetan language), “j” along with “w” (and to a degree “r") are approximants which puts them in a subset of the resonant class. While proponents of multi-vocalic triangular vowels systems underlying Old Chinese might claim that a vowelless analysis, at least at a broad level, just involves pulling “j” and “w” (or the palatal and labial features) out of the vowels (e.g. using “ja” for “e” or “wa” for “o"), it really reflects the fact that only the approximant class of resonants can form a syllabic base in Old Chinese. That is to say “pjt” (which could be over-transcribed as “pjət” or “pəjt” both corresponding roughly to “pit") is allowed, but “pnt” is not. While such an analysis may seem only relevant to Indo-European and excessive for Old Chinese, when one takes a look at Old Chinese rhymes and the earliest Chinese inscriptions (that fed the xiesheng/phonetic series), it becomes a necessity.

    In terms of rhymes, as I mentioned in an older thread, multi-vowel systems for Old Chinese ignore the fact that -an and -en (-jan) can rhyme whereas -ang and -eng (the latter as -angj or -jang depending on where one rather arbitrarily writes the palatal feature) cannot. Statistical evidence is applied to show that actually -an and -en should not be in the same rhyme group because -en (-jan) has a tendency to rhyme more with other -en words than -an. The problem with this is that (as we have seen with the struggles in machine translation and universal grammar) language is not mathematics. The fact than -an still rhymes with -en cannot be swept under the rug as “statistically insignificant". The reason that -an rhymes with -en, but -ang does not rhyme with -eng is because velar codas have a very strong affinity for palatal and labial co-articulations such that -angj with its palatal coda now has a different coda from -ang and cannot rhyme anymore, unlike -an and -jan which can. While one could still write -an and -en and just accept that they can rhyme because underlyingly -en is really -jan, whereas the underlying -jang form of -eng is actually -angj so it cannot rhyme with -ang, it all just gets far too complicated due to an overly phonemic approach obsessed with the sancrosanctity of maintaining a triangular vowel system at even the most underlying level.

    In terms of Chinese orthography (just in case anyone was wondering that this was straying too far from the original topic!), the biggest problem with scholarship on Old Chinese is that on the whole, people writing about the earliest inscriptions aren’t specialists in reconstructing the phonology, and people reconstructing the phonology don’t know enough about the inscriptions. The graphic forms and usages of characters in the inscriptions point to numerous connections within xiesheng (phonetic) series and across related xiesheng series (i.e. same etymology but written with a different phonetic) that are revealed by a schwa/a analysis but ignored when one rigidly enforces a selection of different phonemic vowels. To cite just three different kinds of example (the examples are of course numerous): 孕 is phonetic in 盈 (in reduced form at the top) via a schwa/a ablaut (the phonetic in 孕 itself is shown in the inscriptions actually to be 身 which also supports reconstructing 身 with an original aspirated lateral initial hl- that later converged with reflexes of hn- with which it has been erroneously reconstructed); the word 新 is related to 生 via a schwa/a ablaut (the semantic connection of ‘new’ and ‘birth’ being apparent); the two readings of 覃 are related via a schwa/a ablaut.

    Apologies for the long posting!

  42. J. M. Unger said,

    May 18, 2016 @ 1:39 pm

    Reading Mullaney's piece, I see that he assumes the user can input Mandarin in pinyin using a keyboard (whether QWERTY or Zhai's clever alternative). That is not a trivial requirement even for some Mandarin L1 speakers; it is certainly asking a lot for many Mandarin L2 speakers, including many ethnic Chinese. Mullaney also doesn't take proofreading and error correction into account, though these time-consuming tasks are unavoidable whether keyboarding, doing voice input, or tablet-inscription of characters. The mere appearance of characters in displays or print-outs that people read is only part of what it means to say they are "using characters." A lot of what people who "write" characters today are doing is psycholinguistically quite different from what they would do if they had nothing but paper and pen at their disposal, and most couldn't do input at all without pinyin. So what is Mullaney's point? He seems to be missing the forest for the characters.

  43. JS said,

    May 20, 2016 @ 1:36 pm

    @Chris Button
    Thanks so much for your careful replies. There's not much sense in trying to respond exhaustively to the many issues you raise. My interest is just in good (i.e., powerful) solutions for OC, and I thus try to avoid prior theoretical commitments. A couple points: the apparent ability of a ə/a framework, by limiting the variability of reconstructed OC syllables as regards main vowel, to account for particular graphic connections otherwise unexplained must be measured against the size of what is necessarily the same framework's forest of new implications regarding graphic connections which do not in fact occur. Also: what is schwa/a ablaut? You seem at some points to imply a morphological process of some kind (新/生); at others a kind of phonological variation which produced etymological doublets (談~譚); at others simply one flavor of phonological difference which permitted graphic borrowing ("孕" > "盈"). Suffice to say that aside from 談~譚 type pairs, which many modern solutions can simply claim were doublets in a/ə, I think you are on shaky ground. To take just the last example, it's not at all clear to me that "盈" is based on "孕" (or "孕" on "身"), or what it would mean as far as phonology is concerned if it were — we certainly have a good-sized pile of bad ideas about OC phonology that have begun from overstatement of the phonological implications of Graph B being similar to or derivative of Graph A.

  44. Chris Button said,

    May 21, 2016 @ 6:08 pm


    The Old Chinese ə/a ablaut represents the same primordial state-of-affairs as the e/o (or rather ə/a) ablaut of Indo-European. That is of course not to suggest that Old Chinese and Indo-European came from the same linguistic stock, but rather just reflects the fact that the syllable is the basic building block for any language. Stress/accent was most likely the phonological conditioning factor that gave schwa its variant, but any clearly identifiable grammatical role remains obscure (I find Pulleyblank's extrovert/introvert distinction far too limited as it stands to be convincing). As I think you are suggesting, current popular reconstructions of Old Chinese tend to pick and choose examples to support particular hypotheses while ignoring those that do not fit. When reconstructed properly via a schwa/a ablaut and, just as crucially, a proper investigation into the earliest attested forms of the characters (i.e. not just later forms), the apparent "exceptions" start to fall neatly into place. The resulting word-families can then confirm or refute any putative reconstructions that were made on the basis of one or two subjective associations. Regarding 身, it is the head of a xiesheng series of which the characters 孕 and 盈 are members. The 乃 component of 孕 is abbreviated from 身 which is how it appears in the oracle bones (i.e. 身 as phonetic plus 子); 盈 consists of 孕 as phonetic plus 皿.

  45. JS said,

    May 21, 2016 @ 10:56 pm

    @Chris Button
    Graphically, that "孕" first featured "身" as such is doubtful (not to say the characters are not formally similar); that "盈" featured "孕" is plausible but not clearly demonstrated by evidence I'm aware of. More generally, seeing an early character like "孕" as containing a phonetic component at all is entering (Branner's) crypto-phonogram territory, and seeing the (claimed) fact that "盈" features "孕" as necessarily having phonological implications is a typical kind of analytical overreach — consider semantics. But all this is neither here nor there as regards competing views of the OC vowel space. I want to say I'm interested in seeing something complete using ə/a, but after all, these are the same shoals on which, after so many decisive contributions, Pulleyblank was to run aground — not helped by a misguided idea regarding the tiangan/dizhi. My thoughts.

  46. Chris Button said,

    May 22, 2016 @ 2:31 pm


    Actually by mentioning semantics, you have brought up my other gripe with the current state of Old Chinese reconstruction. It should be obvious that the etymology of Chinese characters should not be treated separately from the etymology of the words they represent. Yet, just the very mention of the "you wen" theory is likely to provoke the same staunch defensiveness as the ə/a ablaut provokes. The fact that the "you wen" theory can very easily drift into fanciful and unsubstantiated speculation undoubtedly accounts for its pariah status. This is particularly acute when "you wen" advocates fail to acknowledge that the actual phonetic root of a character has been confused (often even from the very time of its conception) with a similar sounding alternative, or that the phonetic root itself has become deformed and in some cases merged with another phonetic regardless of its pronunciation. However, people ignore its existence to the detriment of the field. Broadly speaking the two sides of the Old Chinese etymological coin could be represented by Axel Schuessler's "ABC Etymological Dictionary of Old Chinese" (2007) on one hand and Todo Akiyasu's "Kanji Gogen Jiten" (1965) on the other. While Todo's dictionary does often veer into fanciful speculation without any external support in terms of universal semantics (to be fair, he was working over half a century ago without the knowledge we have today), his attempt was absolutely along the right lines and still very useful when used with significant caution. Yet, while Schuessler is often cited (granted it is a much more recent work), Todo barely ever gets a mention. I should note that this is by no means intended as a criticism of Schuessler's extremely useful book since character etymology was not its objective nor should it necessarily have been.

    The character 孕 graphically now looks like it should belong to the xiesheng series headed by 乃, but this is the result of graphic corruption and convergence since there is no phonological connection. Once the "correct" phonetic 身 is re-identified in 孕, the etymological origins of the words represented by the characters 身 and 孕 and other words in the series like 盈 (which does not seem to be using the "wrong" phonetic so to speak) can be investigated. In these three words we have an excellent Indo-European parallel (again not because IE and OC were related but just in terms of universal semantic shifts) in the root *kewh- 'swell' from which developed words like "coeliac" (身 depicting an enlarged belly with "body" being the extended sense as has been noted by others elsewhere), French "enceinte" (pregnant), and "accumulate".

    In terms of "something complete using ə/a", I am actually working on it in my spare time. However, as I no longer work in the field and unfortunately do need to feed myself and family, it is taking quite a while! Finally to your last point, I certainly agree that Pulleyblank's revised proposals for OC initials based on the tiangan dizhi were misled, but I would not discount the value of much of the evidence he marshaled in support of his arguments. The recent proposal for initial uvulars seems to me to be just as misled as Pulleyblank's labio-palatal velars, yet I would not discount the value of the work that led to such a proposal.

  47. JS said,

    May 23, 2016 @ 1:19 am

    @Chris Button
    I'm more or less aware of the epigraphical history of the characters at issue; it seems we just disagree about the matter of "phonetic" esp. as it pertains to "孕", which is fine.

    My impression is different than yours regarding the status of "you wen" — while not a theory as such, I think most would acknowledge that observations of this kind can often lead to good ideas. It is true that Schuessler (2007) is overly cautious in avoiding graphically motivated lines of argument, though probably for good reason.

    That you're no longer a full-timer in the field is all our loss. I wish you the very best in all your endeavors, both mind- and family-feeding!!


  48. liuyao said,

    May 23, 2016 @ 2:49 pm

    While I can still comment, let me tell one character story, perhaps the only experience I still remember from my years of learning characters. (Fortunately I didn't have, or do not remember, any traumatic experience.)

    I was maybe in second or third grade, and the teacher told us this story. Qin Shi Huang had unified China, and one day he wanted the change the character for Qin, which depicted two kings on top of a tree/wood: 琹, clearly a bad sign. His prime minister Li Si 李斯 proposed to invent a new character, taking from the top part of 春 (Spring) and the left part of 秋 (Autumn), and combine to form 秦, in reference to the Spring and Autumn Annals. (Li Si also proposed a new title 皇帝, matching the esteem of the three 皇 and five 帝 of high antiquity.)

    Scholars may well dispute the authenticity of such an anecdote, but as a kid I found it very interesting. I also could see that 琹 looks very similar to the character 琴 (a Chinese zither), also pronounced qín.

    The point is, characters may lead us to the wrong etymology for certain words (though one could make the same claim about wēijī without characters), that's only because they invite us to look more into the word, beyond the sound itself, and that's oftentimes fascinating. I would never have guessed the original meaning of toilet or goodbye, which I only learned about here.

  49. January First-of-May said,

    May 23, 2016 @ 6:24 pm

    A bit of a note: this comment was started on the 15th of May, and relates to the discussion as it was back then (in particular, it has very little on Chinese as such); as I was getting quite close to finishing it (or so I thought), I was told, very sternly, not to write that sort of long comments during exams.
    Yet I'm posting this now (and not after the exams, which won't end until Friday), because I suspect the comment section will close very soon (despite the ongoing discussion).
    With that in mind, the comment follows below:

    @Chris C. (and the rest of the discussion)…
    ""the innovation of representing consonants and vowels with separate symbols was an Indo-European " — I'm glad you picked up on what I took the trouble to imply."

    I'm not really a major linguist (I barely rank up to "amateur", if that), but I want to comment anyway.

    Best we can figure out (it's a bit confusing because the Semitic-based systems spread so well), the "natural" choice for a writing system, aside from "ideographic" (or perhaps a natural development from such), is a syllabary – as in Cherokee, Yi, Japanese (though that was also due to Chinese influence), Linear A/B/C, and to an extent also Sumerian, Mayan, and Hieroglyphic Luwian; and many of the yet-undeciphered systems are also suspected to be syllabaries. (Perhaps even Old Chinese; hard to say.)

    As it happens, in Semitic (Afro-Asiatic) languages, due to the way the grammar works, it makes sense to use a consonant-based system instead of a straight syllabary; this happened with Egyptian, Ugaritic, and the so-called Proto-Canaanite (on which most of the later systems were based).
    This is not the case in Indo-European languages (nor in Sino-Tibetan ones, as far as I know), so syllabaries make sense – but are complicated, and don't work well with phonemes that didn't exist when said syllabaries were made. (It's a good deal easier to add new phonemes – for use in loanwords, for example – to a system that actually records the sounds than to a syllabary, especially a big syllabary.)

    It so happened that, in Greek, the pre-existing syllabaries (Linear B, Akkadian? – it definitely existed at the time, but I'm not sure if it was ever considered for writing Greek) were such a poor match for the phonetics that adopting the Semitic consonant-based system was considered better.
    But since the consonant/(semi)vowel duality of the original Semitic languages was not a good fit for Greek, the respective sings had to be added to for better representation (for a while, early on, different dialects made the choice differently – the Latin letter H was adopted from a dialect that used this form for the consonant, while the dialect that ultimately became standard Greek used the same form for the respective vowel).
    Incidentally, a similar thing happened later in Latin (the differentiation between V and U).

    It wasn't even the only option, anyway. The Aramaic(?) innovation of denoting vowels by diactritics on consonants had worked just fine for assorted Indo-Iranian languages (and even worked its way into some non-IE writing systems, including most of the Tibetan side of Sino-Tibetan).
    Apparently, as that inheritance line got all the way to 'Phags-pa script in Mongolia, vowels became letters on their own, again creating a proper alphabet. This particular line is perhaps best known for being the apparent ancestor of Korean Hangul (which is an alphabet that is trying its best to look like a syllabary).

    For what it's worth, Old Persian cuneiform was also an alphabet, IIRC. Not sure if it predated or postdated the Greek alphabet (and of course Old Persian is also an Indo-European language).

    …That's all, I suppose? I'm not really sure how I intended to continue, anyway.

RSS feed for comments on this post