« previous post | next post »

Most people seem to call it "homophonia" (25,000 ghits), but I'm not even sure what that means:   "Homophonia" (7/31/14).

Following this cartoon in Magic Coffee Hair (8/16/12) and Gretchen McCulloch's article, "What's the Difference Between Homophonia, Homophobia, and Homophonophobia?" (8/1/14) in Lexicon Valley, I'll go with homophonophobia (4,310 ghits), despite the fact that it is a forbidding mouthful, as being a more accurate term for what I want to describe:  an extreme, irrational fear of or aversion to words that sound alike.  In this post, we will discuss homophonophobia, particularly as it relates to Japanese, but also touching upon Vietnamese, Korean, and Chinese manifestations of this type of anxiety disorder.

Peter Gordon has a thoughtful review of Minae Mizumura's The Fall of Language in the Age of English.  The subtitle of the review is "Award-winning Japanese novelist Minae Mizumura brings impeccable clarity to issues of identity and disappearing languages in contemporary language politics". Asian Review of Books (1/26/15); reprinted and more easily accessible in Caixin online (1/30/15).

Originally published in 2008 as Nihongo ga horobiru toki: Eigo no seiki no naka de 日本語が亡びるとき―英語の世紀の中で, the English translation by Mari Yoshihara and Juliet Winters Carpenter just appeared in January of this year.  The following is taken from the Columbia University Press description of the book:

…Minae Mizumura acknowledges the value of a universal language in the pursuit of knowledge, yet also embraces the different ways of understanding offered by multiple tongues. She warns against losing this precious diversity.

Universal languages have always played a pivotal role in advancing human societies, Mizumura shows, but in the globalized world of the Internet, English is fast becoming the sole common language of humanity. The process is unstoppable, and striving for total language equality is delusional–and yet, particular kinds of knowledge can be gained only through writings in specific languages.

Mizumura calls these writings "texts" and their ultimate form "literature." Only through literature, and more fundamentally through the diverse languages that give birth to a variety of literatures, can we nurture and enrich humanity. Incorporating her own experiences as a writer and a lover of language, and embedding a parallel history of Japanese, Mizumura offers an intimate look at the phenomena of individual and national expression.

Much of what Mizumura says resonates with the issues raised in our recent posts on the work of John McWhorter:

"McWhorter on the global linguascape of 2115" (1/26/15)

"John McWhorter responds" (1/29/15)

In this post, I'm particularly interested in Mizumura's opposition to the phoneticization of Japanese on the grounds that Japanese has more homophony than Korean or Vietnamese, both of which have already successfully adopted phonetic scripts over morphosyllabic sinograms.

Abstract concepts derived from ideograms are even more prevalent in Korean than in Japanese, yet in North Korea the use of ideograms is prohibited, and even in South Korea they have all but disappeared, replaced in both countries by Hangul, a phonetic script unique to the Korean language. Vietnam, where a still higher proportion of abstract words was originally written in ideograms, switched to the Roman alphabet in the first half of the twentieth century. Even without ideograms, Vietnamese and Korean have continued to function as written languages largely because their pronunciation is more varied and complex than that of Japanese, resulting in fewer homophones.

(The Fall of Language in the Age of English [pp. 193-194])

Before I present the views of specialists on Korean, Vietnamese, Chinese, and Japanese, let it be said that "ideogram" is not a correct designation for hanzi / kanji / hanja / Chinese characters.  See J. Marshall Unger, Ideogram: Chinese Characters and the Myth of Disembodied Meaning (University of Hawai'i Press, 2004) and William C. Hannas, Asia's Orthographic Dilemma (University of Hawai'i Press, 1987).

Now for the views of experts, all of whom are well versed in at least two East Asian languages.

Bob Ramsey:

Grr! This 'ideogram' nonsense again; that zombie idea just won't die, will
it?  And here Columbia UP goes on printing this stuff.

Anyway, I have no idea about Vietnamese, but It's beside the point.  This
whole homophone idea has always been just a red herring, as you well know,
Victor. Ask anyone Japanese to simply read a passage from a written text
to another Japanese — is Mizumura going to claim it's incomprehensible
without seeing the text?

Ross King:

Certainly Vietnamese and Korean preserve more distinctions/features from Middle Chinese than Japanese does, meaning that many sinographic binoms that end up as homophonous in Japanese will _not_ be homophones in Vietnamese and/or Korean, and presumably Vietnamese does better on this score than does Korean (since Vietnamese is more conservative in its sinic readings and has tones too), but whether all of this actually dignifies the argument that Mizumura is trying to make (the usual Japanese exceptionalism) is another question.

Bill Hannas:

I'd like to see a careful, statistical comparison made before venturing an opinion.  Even that is easier said than done, because you'd have to define up front, for example, what constitutes "Korean" and what part of a large Korean dictionary is simply unassimilated Chinese posing as Korean because someone at sometime assigned a Chinese character or even word a Sino-Korean pronunciation by analogy.  Unless you do that for all three languages any comparison will be hopelessly skewed.

That said, if you look only at the number of phonetic features available in J, K, and V, in particular at the number of distinct syllables (including tone), and even more particularly at the number of syllables available for the Sinitic portions of these three languages to extrapolate the likelihood of homophony, it's pretty clear that Japanese should have more, followed by Korean, and then Vietnamese with the least homophony.  See pp. 88-89 of AOD for specifics.

Jim Unger:

Logically speaking, Mizumura is saying

1. that entire writing systems consisting entirely of ideograms actually exist;

2. that the Chinese writing system is one such system;

3. that a critically large number, if not all, abstract concepts in Vietnamese, Korean, and Japanese are derived from Chinese characters;

4. that, if the ratios of abstract concepts derived from Chinese characters to all (abstract?) concepts in Vietnamese, Korean, and Japanese, respectively, be v, k, j, then v > k > j;

5. that the words in Vietnamese and Korean for abstract concepts (at least the ones derived from Chinese characters) exhibit less homophony than do the corresponding words of Japanese;

6. that written communication has not broken down in the first two languages for this reason alone; and

7. that it has not broken down in Japanese because Japanese still use Chinese characters.

Apart from the well-documented falsity of 1 and 2, I note that the process of derivation postulated in 3 is undefined; that the method for identifying and counting different kinds of concepts implicit in 4 is not described; that no procedure for quantifying homophony in 5 is specified; that what would constitute a breakdown of written communication in 6 is left entirely to readers' imaginations; and that, for all these reasons, 7 merely begs the question.

Eric Henry:

I agree that an abundance of homonyms is not a good reason to reject the possibility of a phonetic writing system.

As for the prevalence of homonyms in Vietnamese, my subjective impression is that they are less prevalent there than they are Mandarin, Korean, or Japanese.

The example that occurs to me at the moment is the compound 意義. In Mandarin this is pronounced yì yì—the same sound twice. In Vietnamese the pronunciation is yí nghĩa—the sounds are markedly different.

Another example is 衛 versus 魏. In Mandarin both characters are pronounced "wèi." In Vietnamese, the first is pronounced "vệ" and the second "ngụy."

Examples such as these are very numerous in Vietnamese.

Liam Kelley:

I think that there is more homophony in Japanese. There seems to be another idea here though that homophony would make it difficult for people to understand words correctly if ideograms were not used because they would not know which term was being used. I don't buy this. Vietnamese will often not know what the root words are in an 2-word expression, but they will understand what the expression means.

Steve O'Harrow:

After half a century of studying and teaching both Vietnamese and Chinese, I am willing to attest that Vietnamese homophony is about 50% greater than any dialect of Chinese. One reason is that all Chinese lexical items have a Vietnamese reading, so VNese gets all those homophones for free [except that, since VNese readings are older than Mandarin, VNese can distinguish between a few Mandarin homophones] and VNese also has a full set of Mon-Khmer homophones, to boot. Thus, romanized VNese is proof positive that arguments regarding the inappropriateness of alphabetizing any language based on the question of homophones is pure bunk.

However, the Japanese are past specialists in denial, specious assertions of uniqueness, and various other forms of elegant bunk, so that some Japanese linguist saying that Japanese cannot be alphabetized (due to some perceived uniqueness of things Japanese [race?]) is neither surprising, nor in the least credible.

As far as Chinese goes, the "problem of homophony" in Sinitic languages is one that has been discussed on numerous occasions here on LLog.  See, for example, this comment by john riemann soong and "Homographobia" (9/27/10), which concludes:

My rule of thumb is always this:  if homography were a problem in (more or less) phonetic scripts based on real, spoken languages, then homophony would be a problem in the speech of such languages.

[Thanks to Lane Greene, John Rohsenow, and Michele Thompson]


  1. Anschel Schaffer-Cohen said,

    February 7, 2015 @ 5:04 pm

    You don't seem to have commented on the title of the book. The Japanese title translates roughly to "The Death of Japanese in the Age of English", but the English title is "The Fall of Language…" Do the two versions differ at all in subject matter? Or is this a case of publishers trying to internationalize the appeal of the book without changing its contents?

  2. leoboiko said,

    February 7, 2015 @ 5:26 pm

    I mostly agree with everything, but this tidbit from Ramsey drew my attention:

    > Ask anyone Japanese to simply read a passage from a written text to another Japanese — is Mizumura going to claim it's incomprehensible without seeing the text?

    I think the answer is going to be "depending on the text, yes". Several Japanese students have said to me things like, "I can't follow this talk unless I have the written text in my hand". Of course, Japanese is perfectly intelligible from sound only, like any living language must be, necessarily. But the question is: to what extent is written Japanese the same as speech?

    According to one count, written Japanese has, on average, about twice as much Sinitic vocabulary as spoken Japanese – 41.3% vs. 23.6%, though it varies with genres and formality (Miyaji and Kai, "Nihongogaku" Tokushu Tēma Betsu Fairu 3: Goi). It would be interesting to see "homophony" quantified and compared in writing and speech; but, at any rate, I believe you'll all agree with me that Sino-Japanese (kango) words, derived as they are from non-vernacular Literary Chinese, are more homophonic than native Japanese. Read a kango-heavy, abstract Japanese text aloud, and the level of homophony might well make it too hard to understand.

    So what's happening here isn't that Japanese is a special language that needs morphograms because it has too many homophones. It's the other way around: the use of morphography allows Japanese writing to use more homophones than one normally would. This may be one reason why people think Japanese is especially homophonic: they're thinking of Japanese written text (as distinct from the vernacular), which uses a lot of homophones for the simple reason that it can get away with it.

  3. David Eddyshaw said,

    February 7, 2015 @ 7:18 pm

    leoboiko is spot on (as you'd expect.)

    There are certainly genres of Japanese writing, especially those heavily influenced by the old Kanbun style, which are far from easy to understand without seeing the written form (I cofirmed this abundantly for myself when I foolishly agreed to translate a formidably technical scientific article for a colleague.)

    Although not as extreme, the case is analogous to the incomprehensibility of Classical Chinese when read out.

    Even ordinary Japanese is surprisingly difficult to read if written all in kana, though I suppose you could argue that's mostly a question of unfamiliarity, and part of it is just the absence of spaces between words and suchlike.

    At least, abandoning kanji would not be just a matter of spelling reform (as it were); it would also necessitate some degree of reform of parts of the written language itself. Whether that is itself desirable is another question, though advocates of scrapping kanji are likely to think so, I imagine.

  4. flow said,

    February 7, 2015 @ 8:55 pm

    @leoboiko thank you for writing "depending on the text, yes", and "the use of morphography allows Japanese writing to use more homophones than one normally would", so i don't have too.

    I think there is an earlier article here on LL where i remarked that one can imagine that the more old-fashioned styles of Chinese literature would not have evolved the way they did were it not for some specifics of the Chinese writing system (for which there are extreme examples of short texts consisting of nothing but characters read xi or shi). For Japanese, there's that humorous story about two people talking about 能楽 and 農学 without them realizing they're not on the same page; one could theorize that the existence of these two homophonous terms is in part supported by their being different in the written.

    As for the commenter deploring the use of the term 'ideogram', this is really becoming a pet peeve of some. First, how did it make it into the English version? What is the original term used? If it only was 漢字 throughout, the author is not guilty, and the translator may be forgiven for using one of the terms with the highest currency. What are the alternatives? Tetragrams? That's not even a word. Logograms? Much worse. Sinitic morpho-syllabic characters? Chinese characters? Sinograms? Anything better than that?

    I really think that as long as you know what you're doing, 'ideograph' / 'ideogram' is not such a bad term. After all we don't complain about physicists who explain that they've just 'split an atom' in their collider, something that should be impossible when 'atom' is taken literally. It has more than once been pointed out in this forum that we should be aware of the fallacy to call some usage 'wrong' just because some word had another meaning back in history; somehow this principle is not often generously applied to 'ideogram'. Speaking of which, i want to remark that the Unicode consortium states that they're using 'ideograph' as a convenient term just to make clear what they're talking about, not as an endorsement of how CJK characters function.

    And truly, of all the writing systems in use in the modern world, if there's anything coming close to the notion of 'ideographs' at all, it's certainly digits, airport pictograms (both with very limited semantic fields, and not full writing systems), and the peculiar way that characters are used specifically in the Japanese language.

  5. leoboiko said,

    February 7, 2015 @ 9:34 pm

    Well, I'm not sure if I'd go as far as calling it a reform of the language itself—at least not to the degree of the Genbun-Itchi reforms; current Japanese writing isn't that far from speech. But I do think that writing in kana or alphabet would definitely imply different vocabulary choices. These should come naturally, if people tried to communicate without relying on kanji.

    A thornier problem in a reform scenario might be the conversion of old texts. Consider literature, as an example of "harder" writing. I'm convinced that Japanese phonographic literature is perfectly possible (we have the Heian period as proof). But if a piece of literature was written, from the start, assuming that kanji are there to draw nuances, then a simple conversion to phonography would make it poorer. If only a handful of specialists can read kanji, only a handful of specialists will get the full effect from Natsume Sōseki or Yukio Mishima. Is that a price worth paying? Or should we, for the love of literary wordplay, keep subjecting children to a much longer route to literacy?

    A clarification: when I said this:

    Several Japanese students have said to me things like, "I can't follow this talk unless I have the written text in my hand".

    I didn't mean foreigners learning Japanese as L2. I was thinking of actual native Japanese college students, trying to follow an academic presentation in Japanese. Contrary to what I was taught in the West about good presentations, it seems that over there professors often stick to reading out loud a pre-prepared speech; and it's this sort of speech that, I was told, is too hard to follow without the handout. I think the professors aren't even trying to make it otherwise… Come to think of, "write your speeches in pure kana/rōmaji" would probably be a good exercise to make them clearer.

  6. Chris Kern said,

    February 7, 2015 @ 9:57 pm

    I have both attended and presented at Japanese academic conferences, but the attendees don't have the text of your speech. As leoboiko says, you typically simply read your paper out loud. Attendees will have a packet or sheet that contains some of the quotations in your paper (particularly from the primary sources), but other visual aids are typically not used. When I gave my presentation, I did not define or explain many of the quite technical words used in the paper, nor were they on the handout. This was apparently the accepted way to do things, and I think it's more of an in-group/out-group cultural aspect — if you don't understand words like 古注釈 then you aren't the in-group that should follow the paper.

    As for literature, Japanese people are already cut off from the literature prior to about 1900, partly due to the non-vernacular writing, partly due to the writing system.

    >assuming that kanji are there to draw nuances
    This is a big assumption. I have never seen any real proof of the common statement that literary authors frequently play with kanji or use them in ways that would be lost in kana or romaji. Audio books exist in Japan.

    Finally, flow:
    I think the difference between "atom" and "ideograph" is that most people understand that atoms are composed of parts, regardless of the etymology of the term. Most people do not understand how Chinese characters work, and are confused at least in part by the terminology used.

  7. J.W. Brewer said,

    February 7, 2015 @ 11:29 pm

    I too am curious whether Mizumura just used the word "kanji" and his translators made the decision to render it as "ideogram," or whether he used some more technical/abstract term in Japanese (that might or might not come with "baggage," the way "ideogram" does here). leoboiko's "morphogram" seems appealing at first glance, but then encounters difficulty precisely because of the distinctiveness of kanji in Japanese, since most kanji represent (via kun v. on readings) multiple morphemes. In my idiolect, I just treat "kanji" as the standard English word and thus think to myself, e.g., "hanja is the Korean word for 'kanji.'" I suppose political considerations might make this approach imprudent for the scholarly community at large.

  8. Peter Evans said,

    February 8, 2015 @ 10:38 am

    The book sounds terrible; I only wish that Roy Andrew Miller were still around to review it. (Though I'd settle for a digested read: http://www.theguardian.com/books/series/digestedread .)

  9. James Unger said,

    February 8, 2015 @ 10:47 am

    One mustn't shift the blame to the translator. Since Mizumura clearly states that ideograms are no longer in Korea or Vietnam, it is very likely that "ideogram" is the translators' choice for _kanzi_, but Mizumura could not assert that K, J, V abstract concepts are derived from _kanzi_ unless she believed that _kanzi_ are (inherently, permanently, and translinguistically) ideograms. (And if Mizumura actualy wrote _hyooi mozi_ where the translation has ideogram, the translators were just being faithful.)

    Steve is right if too polite: it's all "elegant bunk."

  10. leoboiko said,

    February 8, 2015 @ 12:35 pm

    @Chris Kern:

    re: presentations: I can't claim my sample is representative or anything, having been in just a handful of speeches, but in those handful, I was given the full text with the contents of the talk as a hand-out (or, at one time, the speaker just projected the entire text in a large font, slide by slide, and proceeded to read it). I was curious about the fact that native students were looking at the text, rather than at the speaker, so I asked about it.

    re: pre–1900 literature: It's true that most Japanese can't do Classical fluently, despite it being a part of the education system. But the fact that it's a part of the education system tells us that they're trying to avoid cutting the population off, which I find to be laudable. Besides, most classical works are published in editions loaded with footnotes (well, header-notes) and parallel translations (gendaigoyaku); the literary-minded Japanese I know all read those, referring to the modern rendition whenever something is unclear. I even know professional translators who are, er, not quite expert in Classical, but who have published Western translations based on these annotated editions.

    And, if I compare the Japanese reception of their classics with the Brazilian reception of our classics, it's apparent to me that Japanese classics are much more alive in everyday culture. We don't have dozens of annotated editions found in any bookstore, we don't have literally tens of comic-book adaptations of every single famous title, nor tourist reenactions of famous trips, nor erotic fan drawings of medieval poets (yes, I've found Ariwara-no-Nahirina gay porn on the Internet (don't ask)). You may think this thing is ridiculous, and I concur, but the fact that it exists is a testament to the aliveness of classical literature.

    Gendaigoyaku is an interesting phenomenon, by the way. These renderings into modern Japanese lack all grace and literary value, and cumbersomely add stuff that should be implicit, or turn subtle tones into explicit expressions ("I'm being ironic now"). When you look at the way it's presented, side-by-side with Classical (well, top-by-bottom), it becomes clear that they aren't intended to be stand-alone translations, but to be a kind of follow-along explanation to help in reading the original.

    re: Kanji in literature:

    Some examples from the top of my head:

    The most common way of writing hotaru, "firefly", is with the Chinese character for the insect, ‹蛍›. In the novel Hotaru no Haka, Grave of the Fireflies, the author chose instead to use an etymologically transparent orthography, ‹火垂る›. Without changing the sound of hotaru, this leads the reader to analyze it morphologically as “drops” 垂 of “fire” 火. The novel is about the firebombing of Tokyo – fire dropping from the sky. The fireflies are said to be the souls of those who died in the fires.

    I was just working with Yukio Mishima’s Fuman na Onnatachi, about a Japanese tourist who dates a Japanese immigrant in Brazil. After a reluctant, existentially angsty period of courtship, they finally proceed to carnal love in a Rio de Janeiro hotel, right during the Carnaval. "Carnaval" is written in the text as a furigana gloss to 肉謝祭, "flesh-forgive-festival", which is a good reflex of its etymology, informing the Japanese reader of what does it mean. Of course, the author could achieve the same effect through many other means ("The 'Carnaval', which means 'festival of the flesh'…"). But the point is that he didn't; he choose morphography as a way of informing the reader of this. The narrator goes on, saying that the love of their Japanese bodies somehow doesn't feel right in this tropical feast. "Japanese bodies" is 日本人同士の肉体. In a version that's just a thoughtless romanization, the Japanese reader wouldn't spot the relationship between Sinitic niku- and Latin carna-.

    A similar trick is used in Haruki Murakami's Sidney no Green Street. The narrator starts by describing the eponymous street: it's dirty, ugly, depressing, not a speck of plant life anywhere. Throughout this prologue he writes the name of the city in katakana, as ‹グリーンストリート›. At a certain point he says, exasperated, "I can't imagine why did they call it 'Green Street'". At this passage, for the first time, the name of the street is written as a furigana gloss to 緑通り, informing the Japanese reader of what the name means. I call those reverse-glosses – rather than the furigana helping read unknown kanji, it's the kanji that helps understanding the semantics of an unknown word. Again, there are countless ways of doing this without kanji (footnotes, parenthesis, paraphrases), but the point is that this is how the authors chose to do it. They have available a parallel device, allowing them to juxtapose morphography to phonography – morphemes go below, phonemes go above – therefore they make use of it.

    Sometimes the simple choice of kanji is enough to draw nuances. In Bashō's famous poem about cicadas, the word shizukasa "quietness" is written with 閑, rather than the more common 静. From its Chinese use, 閑 suggests "seclusion", and the quietness in question is from a temple high up in the mountains, which Bashō found with the doors closed. Bashō was a fluent reader of Literary Chinese and would understand the implications. Kurahashi Yumiko's Pandora no Tsubo a feminist diss of Greek myth, portrays the Pandora story as ridiculously sexist. She says that the misfortune in Pandora's jar was a misfortune specific to women, "that is, 'netami' [envy]". In this definition between quotes, the word netami is written with 嫉 ­– "women's disease" – rather than the more common 妬. She reverts to 妬 later in the text, which shows that the choice of 嫉 was deliberate.

    Sometimes the choice of writing system is meaningful. In one Jun'ichirō Tanizaki story I can't for the life of me recall the title right now, there are two characters, an educated man and a woman, and the change in point of view is denoted by the difference in kanji density. Often kana is used to denote non-understanding (The Yotsuba& manga does this a lot). Consider this passage from Masayo Koike's Genjitsu-sō, where the unenthusiastic narrator is talking with her friend, a newly mother, over the telephone:



    Normally they do what’s called a tēōsekkai ‹テーオーセッカイ›, you know, where they cut your belly. Yoriko-san told me this as if it was enough for her to be proud of.

    Hmm, tēōsekkai? I’m not as healthy as Yoriko-san, so if it was I giving birth, surely it would have to be that way, right? While I replied Yoriko-san thus, I converted that sound tēō inside my head to the kanji ‹帝王› [“Emperor”], and wondered why they gave it such an impressive name.

    Sometimes the punning isn't related to meaning, it's just playing. Sōseki is famous for this, writing e.g. tonikaku as 兎に角, or hidoi as 非道い. Many of his coinages were later enshrined in dictionaries. Sometimes, rarely, writers even refer to kanji components. This isn't anywhere near as extensive or æsthetically important as Fenollosa though—poems are still primarily about sounds—but sometimes it did happen. There was a Bashō haiku with this sort of component-play, though I can't find it right now.

    This isn't a resource just used in high literature, either. I challenge you to hand me any edition of the Shōnen Jump! manga anthology, and I guarantee I'll find some sort of kanji play somewhere in there. These range from semantic glosses to English and other languages (suupaa as a gloss to 超; "スクレリング" < skræling as a gloss to 土人 (Vinland Saga)); to homophony puns (Ansatsu Kyōshitsu writing -koro "period" as 殺, or yaru "to do" as 殺る); to parallel writing telling the reader what was unsaid (oretachi as a gloss to 赤報隊 (the name of a Meiji-era citizen's army) (Rurōni Kenshin)); to miscellaneous phono-semantic puns (Saint Seiya written as a gloss to 聖闘士星矢 – 聖闘士 seitōshi is "Holy Warrior", and the 'saints' of the series are holy warriors; the "companion demons" of Shin Megami Tensei are called 仲魔 Nakama). The great frequency of kanji play in manga and videogames is interesting to me because most of this stuff is target at teenagers – that is, to speakers still short of full kanji literacy.

    And neither is this kind of play restricted to modern literature; cf. Ariga, The Playful Gloss. Rubi in Japanese Literature and Seeley, A history of Japanese literature, on the part about gisho in the Man'yōshū. Cf. also my master's dissertation in a year or so, if you can read Portuguese :)

  11. flow said,

    February 8, 2015 @ 1:28 pm

    @Jim Unger: you're claiming that Mizumura says "1. that entire writing systems consisting entirely of ideograms actually exist; 2. that the Chinese writing system is one such system; 3. that a critically large number, if not all, abstract concepts in Vietnamese, Korean, and Japanese are derived from Chinese characters;"

    I'm not sure where these statements can be derived from the excerpts provided here and on Caixin.

    Actually, Mizumura does explicitly *not* appear to support #3, "if not all abstract concepts" when she writes "Abstract concepts derived from ideograms are even more prevalent in Korean than in Japanese". If Korean has a higher prevalence of abstract concepts transported first by the use of Kanji, it follows that Japanese must have either fewer abstract concepts in the language throughout, or less of these concepts that have been transported through the use of Kanji. This has nothing to do with whether one endorses the concept of "abstract concept transported through Kanji", but is a matter of logic.

    As for #1 and #2, I can again not see where that view is expressed by the author, or strongly implied. Maybe a suitable quote could help. Other than that, i think what she says is that there is a purely 'graphical' mode of expression, a mode where (to recycle my own example) 農学 and 能楽 and many other pairs (and n-tuples) are 'effortlessly' kept distinct by means of a writing system that, be it 'ideographic' or not, does certainly have a higher degree of 'semantics' going into the orthography than most other modern systems.

    As for the role of Kanji in modern Japanese writing, i think a comparison with Indo-Arabic numerals as used in English gives some hints. First, the numerals are in no way a complete writing system, yet their use in texts often greatly helps to parse structure and meaning. Writing "nineteenhundredeightysix" with or without spaces instead of "1986" feels cumbersome and it is. Likewise, Kanji can help to make a text clearer (although it is not difficult to use them to the contrary effect). Anyways, this mode of expression does *not* rely on the characters to form 'a writing system consisting entirely of ideograms'—all you need is hundreds or a few thousand of squiggles each of which with a reasonably tight association with a reasonably intelligible range of meanings, which is exactly what Chinese characters are, especially as used in Japanese. That these squiggles often do have a tell-tale internal structure that both aids in their memorization and helps to explain how this writing system came about historically is interesting and important, but not a necessary ingredient to the equation as such; it may put an upper limit to what people are able or ready to memorize, but given the complexities and irregularities of modern Japanese Kanji usage, it is questionable how much native writers do make active use of the phonetic clues provided by many 形聲字, and how much is really written out and read as 'yet another squiggle', much like English speakers effortlessly parse "123" without giving a thought to these squiggles having been one, two, and three single strokes millennia ago.

    My impression is that this ongoing bemoaning of the purportedly purported value and mode of operation of 漢字 / 表意文字 / 'ideograms' is really going into the wrong direction.

    Clearly, saying that "Mizumura could not assert that K, J, V abstract concepts are derived from _kanzi_ unless she believed that _kanzi_ are (inherently, permanently, and translinguistically) ideograms" is going over the top to the point where it gets flat wrong. "表意文字" can be be understood as "show-meaning-writing-sign", and this is exactly what the compound says. This way of writing is much more sensitive to the semantics of the morphemes than, say, "hyouimoji" or "ヒョウイモジ", however unique or ambiguous the sounds so expressed may be in a given context. 'Ideograph' / 'ideogram' may be somewhat misleading monikers, but even in case they're not fully extensionally correct (coincide with demonstrable properties of the writing system), they certainly do transport a bit of what users of Chinese writing have been intending, what they felt would make for a "good way of putting words to paper", hence they wrote 菓實 for 果實, 鯨魚 for 京魚 and so on, always trying to tack on that elusive 'semantic' squiggle that would turn a "sound sign" into a "semantic sign".

    Actually, it is exactly the weaknesses of the system that we can use to demonstrate that there *is* a modicum of 'semantic writing' at work here; consider again the characters '表意文字'. 表 is probably thoroughly opaque to most users; that it may have indicated a piece of garment once could be guessed from its lower part, but it's not very obvious. No obvious indication of sound and not a reasonably clear picture—just a squiggle such as '5' or '?'. 意 does appear to have a somewhat-working sound indicator, 音 yin / イン, but the Shuowen believes it's "从心从音", a semantic/semantic compound; now, "sound of the mind; meaning" can be called a passable crib but little more: just another squiggle. 文 is, again according to the Shuowen, just what it looks like, "象交文", "depicting crossing lines", a squiggle if you will. No hint that it could mean "writing" ("meaningful squiggles"), or what its reading could be. 字 is the only character with a truly working sound-indicator, 子, which is placed under a "roof" 宀; 子 is purportedly here also used for its semantic value, "infant", thus indicating "kid in a house; to rear, to nurture". How this character came to mean "character" would need more in-depth research, so what we have is—just another squiggle to many non-specialist readers and writers.

    People, what this means is simply: that apart from the truthful and important assertion by one John DeFrancis that Chinese writing is 80% relying on spoken language with its very sounds (to oversimplify) and much less of an 'idea-writing' than some ("chance in crisis because 危機"; "learning Chinese by pictures") would have it, it's still a writing system that puts a *lot* of semantics into its squiggles, and yes, it does so across languages.

    The Chinese may have failed to create the ultimate way of 'pure idea-writing' that European authors have dreamed up and labelled 'ideographic'. However, denying that 鉄筋 "iron-sinews; reinforcement steel" transports more analyzable, helpful componential semantics than てっきん and 철근 do can not be denied.

  12. Matt said,

    February 8, 2015 @ 6:04 pm

    I've come to the conclusion that word-level homophones are a red herring. The kanji system (restricting myself to Japanese only) is best thought of as a mnemonic tradeoff: accept the considerable burden of learning thousands of kanji, and in return you get to store part of your vocabulary outside your brain. The problem isn't for example that on seeing "sōshi" you would be unable to guess which of the 20+ meanings are intended — it's that many of those rarer meanings wouldn't even occur to you unless you racked your brain for a while, which is obviously not a fun way to read. Kanji jog the memory. You can see something similar to this at work in modern editions of Heian kana texts that gloss strings of kana with kanji. Most of the time the kanji isn't particularly surprising; indeed it's the only one that would make sense in context. But it's easier and faster for the average reader to see it than think of it themselves.

    Of course it would be possible to do this stuff in other ways, for example with parenthetical explanations like those used in speech. But even though it would clearly be possible to write a form of Japanese phonemically, the burden of proof for the idea that this change would not have a net negative effect on the literary culture is on the pro-kana side, not the kanji conservationists'.

  13. leoboiko said,

    February 8, 2015 @ 6:37 pm

    Sorry about countless typos in that wall of text! The Seeley reference was actually A History of Writing in Japan, not "of literature".

  14. Victor Mair said,

    February 9, 2015 @ 12:16 am

    @Anschel Schaffer-Cohen

    Publishers of translated books often make adjustments in their titles. For example, Bantam asked me for dozens of different suggestions of titles for my translation of the Zhuang Zi before they were satisfied with Wandering on the Way, which I myself like very much. In the present case, it is clear that they were trying to broaden the scope and potential readership of Mizumura's book.


    I am grateful for all of your informative, insightful, and fair comments, and your lack of tendentiousness and snideness is particularly refreshing. There is much that we can learn from your remarks. Amidst all of your other carefully considered sentences, I especially appreciate ones like this: "So what's happening here isn't that Japanese is a special language that needs morphograms because it has too many homophones. It's the other way around: the use of morphography allows Japanese writing to use more homophones than one normally would."

    @David Eddyshaw

    "At least, abandoning kanji would not be just a matter of spelling reform (as it were); it would also necessitate some degree of reform of parts of the written language itself."

    I'm not sure I understand what kind of "spelling" reform and what type of "reform of parts of the written language itself" you're referring to. If you're talking about a reform of written style, so that what people write is closer to the way they talk, that has long been a desideratum of language reformers in China and elsewhere.

    Your next sentence ("Whether that is itself desirable is another question, though advocates of scrapping kanji are likely to think so, I imagine.") is gratuitous. Your use of "likely" and "imagine" reveals that you're just speculating on what "advocates of scrapping kanji" might think. And just who are these "advocates of scrapping kanji" anyway?

    @J.W. Brewer

    Minae Mizumura is a woman.


    To question the use of "ideogram" is not a peeve. It is a matter of whether or not it is accurate to refer to the hanzi / kanji / hanja by that term. Your own comments are fraught with waffling, reservations, and qualifications about the aptness of the term. Most people just call them "Chinese characters", and I'm happy with that, although when writing for an audience of linguists there are other terms that more accurately describe their nature and function.

    From a colleague (on whether it was Mizumura or the translators who chose the word "ideogram"):


    I don't think it matters. Mizumura may have written 漢字, but she certainly could not assert that they are the source of (at least some, if not all) abstract concepts in Korean, Japanese, and Vietnamese (from which they were allegedly "derived") unless she believes that they are, permanently and inherently, 表意文字.

    To put it another way, either she wrote 表意文字 in every place the translator writes ideogram, in which case the foregoing interpretation is empirically justified, or she wrote 漢字 in all or some of those places, in which case the translator at least evidently concurs with the foregoing interpretation.


    If you haven't already done so, please read the books by Unger and Hannas that I referred to in the original post.

    Your choice of 鉄筋 as an illustration for the defense of "ideograms" as a superior form of writing is particularly unfortunate. First of all, there are at least ten different ways to write 鉄 (Chinese tiě — trad. 鐵; simpl. 铁), and Chinese teachers of the traditional forms of the characters do not agree on which is the correct one.


    There are at least seven other different ways to write 筋. (Chinese jīn — trad. 筋; unofficial simpl. 竻)


    The traditional form of 鐵 is difficult to write, and it is not immediately obvious upon what principle the simplified form 铁 (Jap. 鉄) is derived (unless one is familiar with historical phonology).

    The jīn 金 semantophore indicates that 鉄 / 鐵 has something to do with "gold" or "metal", but how we get from there to "iron" isn't clear. The zhú 竹 semantophore tells us only that 筋 has something to do with "bamboo", but that doesn't do much to help us get to "muscle; vein; sinew; tendon".

    The phonophores of both 鐵 / 铁 / 鉄 are far from obvious.

    Given the semantic and phonetic opacity of 鉄 and 筋, the learner has no alternative than to sǐbèi 死背 (lit. "die-back", i.e., "memorize without comprehending" — there's a long story in the history of Chinese education about why sǐbèi 死背 has this meaning). Just memorize arbitrarily that 鐵 / 铁 / 鉄 is pronounced tiě in Mandarin, tetsu in Japanese, and cheol in Korean and that 筋 is pronounced jīn in Mandarin, kin in Japanese, and geun in Korean (of course, they have plenty of other sounds in earlier forms of Japanese and in Cantonese, Shanghainese, Hakka, Taiwanese, etc.).

    I submit that far more people know how to say tiějīn in Mandarin, tekkin てっきん in Japanese, and cheolgeun 철근 in Korean and what those words mean than know how to write the hanzi / kanji / hanja for them. Those who do know how to write those words in Chinese characters had to invest an enormous amount of time and energy in endlessly repeating the neuromuscular movements to engrave just those two characters in their memory. Conversely, it only takes a couple of days to learn pinyin, kana, or hangul, after which one can write down anything that one can say and read anything that is written in those phonetic scripts.

    So, what is a tiějīn / tekkin てっきん / cheolgeun 철근? It's a rebar (short for iron reinforcing bar).

    I don't know how many homophones there are for cheolgeun 철근 in Korean to be afraid of, but in Japanese I only know of one for tekkin てっきん, and that means "glockenspiel; carillon", so there's not much danger of getting that mixed up with "rebar".

    In actuality, the word for "rebar" in Mandarin is not tiějīn 鐵/铁筋, but usually gāngjīn 钢筋 and sometimes luówén gāng 螺纹钢, but I'm too tired to talk about that tonight. Maybe some other time.

    @Chris Kern

    Your concrete, practical examples of actual usage, e.g., in presentation, are much appreciated. The note about audio books is very much to the point!

    @Peter Evans

    I agree: it would have been great to have a thorough debunking of Mizumura by Roy Andrew Miller.


    "…even though it would clearly be possible to write a form of Japanese phonemically, the burden of proof for the idea that this change would not have a net negative effect on the literary culture is on the pro-kana side, not the kanji conservationists'."

    The burden of proof is upon neither side; the burden of proof is upon history. What will happen in the future depends upon what the people decide through their actual practice, not upon what the pro-kana/romaji or pro-kanji sides say or do.

  15. Elessorn said,

    February 9, 2015 @ 7:56 am

    Professor Mair, with respect, I ask you to consider the possibility that you're not being entirely fair here.

    1. Almost every participant, including many of your colleagues as quoted, explicitly understood the issue as raised in the context of an ongoing polemic about script reform, and responded in kind. Your intent is not to enter the lists on either side of any such debate, I know, but given the frequency of the apparent misinterpretation by reasonable-seeming people, surely it merits more than dismissal.

    2. You reference "gratuitous" asides and hint at a background of snide-ridden tendentiousness. Yet I ask you: does a good faith discussion with Mizumura happen after the following?

    …whether all of this actually dignifies the argument that Mizumura is trying to make (the usual Japanese exceptionalism) is another question…


    …However, the Japanese are past specialists in denial, specious assertions of uniqueness, and various other forms of elegant bunk, so that some Japanese linguist saying that Japanese cannot be alphabetized (due to some perceived uniqueness of things Japanese [race?]) is neither surprising, nor in the least credible.

    It seems terribly corrosive to assume that behind a writer's clumsy, searching attempts to explain the unique value to herself of her own cultural practices there must lie hidden, somewhere, a belief that these practices are in fact superior to those of others.

    3) Which really gets to the heart of the matter. Why are we discussing what seem like her ad hoc musings like serious claims about Japanese linguistics? To me, good philological practice would suggest that we treat a writer's offhand statements about her language–on which she is no expert, but all the same a valuable informant–as expressions of experience that may, under serious analysis, yield worthwhile insights. At least I can't understand what a "debunking" of a subjective essay might look like. I'm sure she serves up plenty of nonsense about language–linguistics is science, not gut. But reading with something like an annoyed swatting reflex seems a surefire way to miss any points she might have. Reminding me of the fate of many considerations of characters on these pages essayed from a non-utilitarian point of view…only to be criticized for their failure to satisfy a utilitarian critique.

  16. flow said,

    February 9, 2015 @ 8:35 am

    @VHM i basically concur with everything you're saying in your comment, but please let me clarify. First, i do think the choice of terms here is important, and i do not think 'ideogra(ph/m)' is a particularly good choice. Complaints about its deleterious effects are justified. A better word would be good to have, but none of the alternatives has gained much currency so far. Lexicograph? Morphograph? Sinograph? Logograph? Where it looks like it's becoming a peeve is when someone reads it and starts to complain. I certainly am guilty of this when i realized Unicode calls Chinese characters 'ideographs', until i read their disclaimer.

    Second, i do not claim the characters 鉄筋 are particularly easy-to-learn, readily and intuitively available, 'magic' founts of wisdom. "Your choice of 鉄筋 as an illustration for the defense of 'ideograms' as a superior form of writing is particularly unfortunate" is a misrepresentation, i'm afraid, when i write "鉄筋 transports more analyzable, helpful componential semantics than てっきん". I do not think these characters demonstrate a superior form of writing, just one that puts more semantic considerations into the written expression than most other forms, however clumsy and hard-to-learn the outcome may be. Believe me or not but when i see 鉄 i associate tie, tetsu, iron without analyzing the character. When i'm unsure i can sometimes recall the associations by looking at the parts, sometimes guessing the approximate reading is possible, and otherwise i'm at a loss. As for 筋, i'm certainly at a loss to reproduce it correctly out of thin air, to recognize it with certainty out of context, or to derive any meaning oder reading from its componential setup alone.

    The reason i chose 鉄筋 is because i remember that after the Hanshin earthquake, when i found myself glued to the TV trying to catch up with the news, some professor did a Japanese-style lecture in the program explaining why the Hanshin Expressway had collapsed. He did so with a surprising wealth of onomatopoeic expressions, and at one point he referred to some detail of the てっきん having a problem—when the editors decided to add an explanatory subtitle "鉄筋" along with furigana. Although i had not seen or heard the word before, it was immediately obvious what it referred to, and it certainly looked like it was intended to be helpful for the Japanese viewers as well, notwithstanding the considerable complexity of the characters involved. And yes, homophony is not the problem here. To me this use of characters-as-an-explanation was certainly new; my point is that it works where known characters are used in reasonably transparent compounds. What you pointed out as difficult points when trying to guess or learn the two characters 鉄 and 筋 i pointed out using the four characters 表, 意, 文 and 字 as examples. This is not an easy script.

  17. flow said,

    February 9, 2015 @ 8:46 am

    @Elessorn: +1 for "non-utilitarian point[s] of view [are] criticized for their failure to satisfy a utilitarian critique".

  18. Victor Mair said,

    February 9, 2015 @ 10:04 am

    From a Japanese linguist:

    …"spelling reform" or "abolition of characters" are more or less strawmen in this day and age. I think the real issue is whether or not political leadership in China, Taiwan, Korea, and Japan will acknowledge and encourage the natural trend toward digraphia in their respective countries in a constructive way, or resist it in a reactionary way.

  19. Jacob Li said,

    February 9, 2015 @ 11:52 am

    It's funny to see a commenter not being able to tell that "Minae" is a woman from the romanized given name. If one sees the Kanji it's pretty easy.

    That said, an experienced Japanese reader knows that Japanese given names ending with "ae" is likely female, and there are names like Andrea or Ashley in East Asia too. Nevertheless, I feel that telling gender from names is easier in general from Kanji/Chinese characters than romanizations.

    On a side note, as a Chinese currently learning Japanese I find it an interesting experience seeing Japanese authors playing with Kanji homophones (mostly for humor). I recall that Prof. Mair wrote about the "scriptural juxtaposition" in Chinese popular culture. As far as I can tell, it's a practice "imported" from Japanese popular novels that use ruby characters, with a different meaning than the kanji/kana/whatever it annotates, for comical effects.

    In Chinese it is almost always pinyin-on-Chinese because pinyin is pretty much the only "secondary writing system" that is available (in Taiwan it is usually bopomofo). In Japanese it can be more diverse: kana-on-kanji, kanji-on-kana, or kanji-on-kanji (homophones usually play a part in this case). The gist of this trick, from my personal experience, is that the second script is more difficult to understand than the primary one (well, in Japanese the ruby script is in smaller print, and thus naturally more difficult to understand). In this way it creates a delay in understanding the second meaning, hence the delayed delivery of the pun. It's not necessarily an indication that the native speakers find it easier to write/understand in pinyin or bopomofo.

  20. J. W. Brewer said,

    February 9, 2015 @ 2:47 pm

    For me personally, romaji is better than kanji for getting me to predict the sex associated with a Japanese given name, due to my extremely poor level of kanji literacy. Whatever sense of that I have was not good enough to infer correctly from Minae, but since I didn't notice that she was referred to with female pronouns in on one of the English-language block quotes, that doesn't particularly matter. (One of the "tells" I thought I did know is that names ending -ko are highly likely to be female, but I googled up a list of the 50 most common names for girl babies in Japan in a recent year and it was ko-less, so I guess names run through fashion cycles there just as they do in the West.)

    I do think there's something to her critique (as related by Gordon's review) of "phonocentrism," but her claim that it is a "Western ideology" overlooks the rather important fact that the opposite perspective ("graphocentrism"?) is also a "Western ideology" and quite probably the stronger one in most Western intellectual circles outside of marginal places like linguistics departments. That said, the phonocentric perspective does seem to have gotten into the discipline originally via doing anthropological linguistic fieldwork among non-literate peoples in the Trobriand islands or wherever. And I could see that it might be irksome for someone from a society which had had a literate elite for almost as long as France has had one to feel like her society had been slotted into that perspective.

  21. Bob Ramsey said,

    February 9, 2015 @ 3:01 pm

    There've been a lot of interesting and intelligent things said on this chain so far; I was particularly taken by Chris Kern's observation about Japanese academic conferences and the importance of in-group/out-group culture there. We all know what he's talking about, even apart from the Japanese particulars. How often do we see similar things happen in English-language presentations here? I also liked the notes posted by Leoboiko and others about punning and playing with kanji in Japanese culture, particularly (but not only) in literature. Just impressionistically, such word play appears to happen much more in Japan than in, say, Korea.

    It seems to me, though, that much of this discussion has strayed a bit from the original topic. The question Victor started the chain off with was about homophony, and whether or not it presents problems that are uniquely Japanese–or at least so unusually weighty kanji are required to disambiguate, while Chinese characters are not required, say, in Korean.

    Notice that this question is not the same thing as questions about the role of Chinese characters in writing more generally.
    Yes, of course, some written texts cannot be read with complete understanding by the listener. (Goodness knows, that's certainly true of English-language books as well.) The question here, though, is to what extent that opacity is the result of homonym confusion. More to the point, does homonym confusion reach a tipping point in the case of Japanese but does not in Korean or Vietnamese?

    I'd say to anyone claiming that's the case (such as Mizumura), that the claim remains to be proved. We need much more than anecdotes to validate the claim.

    I for one suspect it's not the case, and that instead we are dealing with particularities of Japanese culture more than with any unique characteristics of the language. But, as I said, we need a little more empirical evidence before making a claim.

    A final note about the more general utility of kanji (and not homophony): I like the point "flow" makes about the convenience of characters by comparing them to Indo-Arabic numerals. Koreans often say the same thing. For the most part, you only see Chinese characters in South Korean newspapers these days for (1) on occasion, personal names (homonym disambiguation in this case?) and (2) as headline shorthand, e.g., 日 for 'Japan'; 中 for 'China'; 美 for 'U.S.'; etc. There are plenty of potential homonyms for each of those latter syllables. Even so, note that there would be no confusion at all if Hangul was used instead for those country designations–just as "1986" would not be confused if spelled out in either Hangul or Romanization.

  22. J. W. Brewer said,

    February 9, 2015 @ 4:28 pm

    I think the argument would run like this — Japanese phonotactics give it a low number of available "legal" syllables compared to many languages (including the comparators here). It avoids excessive homophony in its "native" lexicon the obvious and natural way — by having lots and lots of polysyllabic morphemes.* But the large chunk of Sinitic-origin vocabulary is generally comprised of [I did that on purpose b/c of the other thread . . .] monosyllabic morphemes because that's what it was in the source language, and since modern Japanese phonotactics are more restrictive than Middle Chinese phonotactics were (without getting into the issue that source-Chinese pronunciation was not uniform and there are thus different strata identifiable in the loanwords acquired by Japanese) the homophony problem is multiplied beyond what it was in the source language, especially w/o tone to disambiguate. The strategies one would use for disambiguation in a non-written context (most obviously turning monosyllables into polysyllables – the way, e.g., Americans with a pin/pen merger can say "ink-pen" to specify which homophone they mean and one can imagine an endpoint where it was no longer cromulent to just say "pen" for "ink-pen") are less necessary, and thus less likely to be used, if the homophones are distinguished in writing and used disproportionately in a written context. Note that to the extent modern Mandarin has significantly more two-syllable (and two-character) morphemes than Middle Chinese did (which I think I believe to be the case because I've read it on language log?), the same thing may not have happened to the same extent in Japanese in large part because the sinitic-origin lexicon was used in a more specialized way that resisted the disambiguation dynamics that caused that development in Mandarin.

    I don't know that that's all empirically accurate, but it seems perfectly plausible.

    *Example: for Mandarin speakers the imminent Lunar New Year will begin the Year of the Yang2, whereas for Japanese speakers (having abandoned the old lunar calendar during modernization), January 1 began the Year of the Hitsuji.

  23. J. W. Brewer said,

    February 9, 2015 @ 5:04 pm

    Somewhat shorter addendum – I am dubious about Bob Ramsey's dichotomy between "characteristics of the language" and "particularities of Japanese culture" because you cannot separate the culture from the language when part of the culture includes attitudes and behaviors regarding and affecting language use. Most of the obvious historical examples of significant change in writing system (although it was claimed above that that's an off-the-table straw-man . . .) were carried out by political regimes that were fairly obviously trying to change the relevant cultures, and did. That was certainly true in Turkey, and has been true e.g. pretty much everywhere there has been oscillation between Latin and Cyrillic. (You might think the changes were positive in any given instance, but the point is that it's naive to think that the change in writing system was a purely technical/pedagogical improvement not linked to a broader agenda.) In East Asia, it seems to me as a non-specialist that modern Korean and Vietnamese nationalism has involved inter alia a desire to get some distance from the oppressive-feeling historical weight of Chinese cultural influence. Modern Japanese nationalism, finding itself militarily and economically superior to China while still using kanji, did not feel the same sense of burden or grievance associated with the Sinitic-seeming things Japanese culture had borrowed and then assimilated, and thus did not have the same psychological need for symbolic differentiation.

  24. Elessorn said,

    February 9, 2015 @ 5:55 pm

    @Bob Ramsey

    Notice that this question is not the same thing as questions about the role of Chinese characters in writing more generally. … The question here, though, is to what extent that opacity is the result of homonym confusion. More to the point, does homonym confusion reach a tipping point in the case of Japanese but does not in Korean or Vietnamese?

    I agree this is a more productive way of approaching the question, though I feel it's quite clear that the discussion from the beginning was framed, and conducted, precisely as if the role of Chinese characters in general was the issue at hand. Need not have been, of course, but was.

    Re: what are being taken as Mizumura's claims, it's hard to say until I can get my hands on the book, but my hunch is that homonymity in its customary English usage- "boar" vs. "bore" for example- is not likely what she had in mind. Not to say that there aren't officially tons of such homonyms floating around– as was pointed out, the phonetic reductions of Japanese basically assure this is the case to a greater extent than in Korean– but rather that I think most Japanophones upon hearing such claims would interpret them not in their strict sense, but instead as a gesture to the very, very familiar experience of not immediately intuiting from context alone what kanji was being intended by a word (or part of a word) as it hits their ear. Memory immediately lights up with the association of kanji and grateful clarification, learned from countless occasions of momentary, often minor, confusions dispelled immediately by finding out, or asking for, the characters involved.

    A (clearly new) way people sometimes phrase this experience is that they "can't convert" (henkan dekinai 変換できない) a confusing segment of speech. This is a metaphor borrowed from computer (or let's face it, cell phone) text input methods, which are experienced by Japanese users as a constant stream of "character conversion" (moji henkan 文字変換) choices floating along the bottom of the screen. Everyone is familiar with the frustration of–for some reason–not getting the desired characters displayed immediately, and I think it's a perfect metaphor for the ways kanji confusion is felt in daily life. The same things that jam up "smart" conversion software– abbreviations, neologisms, puns, listener misparsings, interlocutor knowledge gaps– trip up speakers in real life as well. And while it's true that character homonymity is at the ultimate root of a lot these communication malfunctions, it's not really what we mean by saying "homonym confusion."

    I think if you were to ask Japanese people whether they are often stopped short by identical sounding word-pairs, they would say no, though it definitely happens more frequently than in English. But if you were to ask if they've had the experience of not fully understanding a text read out loud without seeing the characters, they might well say: all the time.

  25. Elessorn said,

    February 9, 2015 @ 6:28 pm

    @Victor Mair

    …"spelling reform" or "abolition of characters" are more or less strawmen in this day and age. I think the real issue is whether or not political leadership in China, Taiwan, Korea, and Japan will acknowledge and encourage the natural trend toward digraphia in their respective countries in a constructive way, or resist it in a reactionary way.

    No one has given any indication of disagreeing with this view, though "natural" should really be "observed". At least in this discussion I can't see anyone likely to dispute your statement that the preferences of users will and should decide how things turn out.

    But straw-man? Presumably this linguist, and yourself, would not agree then that there is something unsettling in the also easily observable phenomenon of Western non-users or guest-users of character systems having weirdly strong opinions about the obvious necessity of those systems being reformed? It seems a very live issue to me that such an attitude comes naturally even to a significant minority. One doubts it tends to correlate with the humility necessary for true appreciation of a foreign culture, and one doubts that this kind of humility is going to become less important in the globalizing future– quite the opposite. But this really is digressing, so I'll stop here.

  26. Bob Ramsey said,

    February 9, 2015 @ 8:38 pm

    "[M]y hunch is that homonymity in its customary English usage- "boar" vs. "bore" for example- is not likely what she [Mizumura] had in mind."

    I think you're probably right. But if so (and I think it is), I'm not sure her argument vis-a-vis Korean holds water. I'm old enough to remember a time in Korea when the uncertainty or confusion about a textual interpretation you describe as: "…not immediately intuiting from context alone what kanji was being intended by a word (or part of a word) as it hits their ear" very much applied in Korean as well. Forty-some-odd years ago, when I first dealt with professional writings in Korean, I recall many occasions when Korean scholars needed to see the written version of a text to understand it completely. But we find today, when Chinese characters have completely disappeared from most such texts, scholars, esp. younger scholars, have no trouble at all maintaining scholarship at an even higher level with little or no such difficulty. (I confess that I personally still find it easier to read linguistics writings containing characters, but then again, I'm old and not a native speaker…) Koreans have adapted rather easily to the loss of Chinese characters.

    So I repeat my question: are we to believe there's a tipping point of homonymy that Japanese has passed, but Korean has not? I mean, were Koreans all along free to use hangul or whatever phonographic writing they wished, while Japanese were so constrained by the homonomy of their language they did not have that choice–or that such a choice would be intolerably destructive to their culture?

    (Of course in making their choice to stop using Chinese characters, Koreans did make certain cultural adjustments, but we can see now that those adjustments ended up being tolerable and actually pretty minor.)

  27. Eidolon said,

    February 9, 2015 @ 10:44 pm

    Chinese characters are not ideograms, but I have a feeling the writing system does lend itself to a larger degree of semantic conservatism than alphabetic systems. I am a native speaker of English, but I can't read a lick of German; the phonetic changes, however slight, coupled with the differences in orthography to accommodate those changes, make even obviously cognate words difficult to make out unless you knew what to look for. In short, I'd need training to make out the large corpus of cognate vocabulary that surely exists between these two languages, much less the German specific words.

    On the other hand, as an intermediate reader of simplified Chinese characters, and associate to several expert readers of traditional Chinese characters, my experience has been that Japanese orthography, though still not readable, is at times off-and-on comprehensible when there is a high % of characters. I am also certain that, had Japanese orthography been alphabetic, such comprehension would disappear, because the Japanese pronunciation of these terms are sufficiently different as to be alien to speakers of Sinitic languages. Indeed, there'd be a lot less comprehension than between English and German.

    This is not to say, of course, that Japanese uses Chinese characters in the same way; indeed, they have their own characters unique to Japanese, and also use a lot of shared characters in a different way. But there is an intersection, and this intersection allows those who know Chinese characters to read a fraction of Japanese writing. It's not intelligibility – heavens no – but it's also not compete incomprehension. It allows for a degree of inter-lingual communication without actually knowing the spoken language – ala disembodied semantics. It's easy to see, then, why ideograms, however inaccurate a description of Chinese characters, continue to catch on.

  28. Victor Mair said,

    February 10, 2015 @ 1:04 am

    From Jay Rubin:

    The English translation has been extensively adapted from the Japanese original so as to emphasize and amplify the book's substantial cosmopolitan content and make it more relevant for non-Japanese readers. As mentioned in Mizumura's "Preface to the English Edition," Juliet Winters Carpenter took on this daunting task with the author, and their combined efforts have proved extraordinarily successful–which makes it all the more difficult to compare the translation against the original. We can find "ideogram" used for both 表意文字 and 漢字. Certainly both the author and the translator are aware of the questionable nature of the term. On page 106 we find, "Chinese characters are ideograms and not phonograms; that is, they represent meaning and not sound. (More precisely, most Chinese characters are phono-semantic compounds, having a dual function, but I will use the term 'ideogram' to simplify my argument.)" On page 159 of the original, we find the rather different corresponding passage–

    . . . 漢文は漢字で書かれている。漢字は、ローマ字アルファベットなどとはちがい、表意文字である。(正確には音と意味と両方を表すの で「表語文字」だが、ここでは話の運びを単純にするため「表意文字」として扱う。)すなわち、声に出して読むことよりも、目で読むことに 重きを置いて発展してきた文字である。/ Kanbun texts are written with kanji. Unlike the Roman alphabet and such, kanji are ideograms. (More precisely, they are logograms which express both sound and meaning, but here I will use the term "ideogram" to keep the discussion simple.) In other words, they are characters that developed with a greater emphasis on reading with the eye than reading aloud.

    Clearly, the author and translator have collaborated closely. I'm writing a review for TLS that will probably appear sometime in March. It focuses primarily on literary matters, but includes the following paragraph–

    There is so much nonsense circulating about the ineffable mysteries of the Japanese language that it's hard [for the general reader] to know what to believe. That old red herring Mizumura cites about Japanese sentences not having subjects, for example, is a myth. All Japanese sentences have subjects. Otherwise, they wouldn't be sentences. (See my Making Sense of Japanese [2012, Kodansha USA].) And Mizumura's use of the long-discredited term "ideogram" to refer to Chinese characters, the most "magical" element of written Japanese, seems calculated to drive linguists crazy by preserving another myth–that you have to see Chinese and Japanese to understand them.

    Anyway, what's really important is that I've got a novel coming out in May called THE SUN GODS (it's already listed on amazon) that will give people a lot more opportunity to dump on an author than Mizumura deserves for her outmoded terminology.

  29. PeterL said,

    February 10, 2015 @ 1:37 am

    I just ran across this item in my Japanese dictionary:
    外国人登録原票記載事項証明書 [がいこくじんとうろくげんぴょうきさいじこうしょうめいしょ: GAIKOKUJINTOUROKUGENPYOUKISAIJIKOUSHOUMEISHO] certification of information recorded on foreign resident registration file

    Of course it would be more comprehensible if written gaikoku-jin tōroku genpyō kisai jikō shōmēsho (or がいこくじん とうろく げんぴょう きさい じこう しょうめいしょ), although people who can read German might disagree.

    And I wonder how comprehensible it would be if read out-loud? Perhaps that's why NHK news often has the headline written out — to give context to the what the newsreader is saying.

    Homophones are also less common if the mora pitch is taken into account (and which isn't shown with either hiragana or rōmaji), although there is considerable variation amongst the major dialects.

  30. Victor Mair said,

    February 10, 2015 @ 10:25 am

    Regarding the intelligibility of Japanese when it is read aloud, much depends on the manner in which it is vocalized, including pitch, stress, emphasis, pauses, etc. Readers do this naturally as part of the process of making sense of the text. If they just drone on syllable after syllable, they are probably not understanding the text themselves. In other words, reading a text involves a certain amount of grammatical analysis, and that is conveyed through the manner it which it is articulated aloud.

    Even in Literary Sinitic / Classical Chinese, which is supposedly (but not really) monosyllabic, the manner in which a text is read aloud makes a great difference when it comes to understanding what it's all about. I've been teaching Classical Chinese for a long time, and I always have my students read aloud each sentence of a text before they translate and explain it. Just from the rhythm, pauses, and intonation of their reading, I can almost always tell immediately whether they understand what they're reading or whether they have misconstrued the text, and I can often tell them how they have misunderstood the text just from the way they read it out loud.

    For the first few months of a new class, the students are unnerved by how I can tell them right away whether they understand the text just from the way they read it aloud, and but after awhile they get used to it and think hard about how to parse, phrase, and frame their reading before they start to recite.

  31. Elessorn said,

    February 13, 2015 @ 1:27 pm

    @Bob Ramsey

    {This is far too late coming to expect a response back, but in addition to begging your pardon for circumstances causing my sudden withdrawal, out of sheer interest I wished to reply}

    Thank you for the fascinating comment. I especially appreciated the personal account of Korea before the real end of character use. Whether the Japanese retain their three-cornered system indefinitely or opt for the full phonetic plunge, Korea will always be the case of comparison–as it should be. At the same time, I think there are important differences between the two cases. To start with your conclusion:

    So I repeat my question: are we to believe there’s a tipping point of homonymy that Japanese has passed, but Korean has not? I mean, were Koreans all along free to use hangul or whatever phonographic writing they wished, while Japanese were so constrained by the homonomy of their language they did not have that choice–or that such a choice would be intolerably destructive to their culture?

    Re: homonyms– My direct answer would be, it seems very likely to me that the greater average phonetic distinctiveness within Sino-Korean allowed more of it to survive the loss of character mastery than would have been the case otherwise–and that, yes, as a corollary, less would survive a similar transition in Japan. I might avoid the word "tipping-point" because I don't think homonymity alone is a sufficient cause of anything, but I would agree with the rephrasing: "Koreans [were] all along free[r] to use hangul or whatever phonographic writing they wished [without great loss], while Japanese were [more] constrained by the homonomy of their language." I suspect there might even be some clever way to (very roughly) quantify this "homonymity disadvantage quotient," but I'm not sure it would matter too much, because…

    My deeper answer is a variation of what Leo said above: that the extreme (or at least notable) degree of hononymity observed in Japanese is as much a result of kanji's deep-rootedness as it is a cause. Currently, the mental model of Least Common Denominator Audience (LCDA?) in any communicative interaction between native Japanophone adults in any medium is going to assume basic kanji literacy. This expands the space of free play that writers/speakers have to work with and listeners/readers have to deal with. In this sense Matt is totally right about it being a red herring: homonymity is not a reason for needing kanji (nor an argument for keeping them), but an index to the depth of their influence on the language. It serves as an example of how the use of kanji is descriptive of Japanese in a way that neither the alphabet is of English, not even the Arabic script was of Ottoman Turkish.

    And so of course while not literally "intolerably destructive of their culture" (cultures elsewhere have clearly survived much worse catastrophes than language reform), Geoffrey Lewis' immortal characterization of the Ottoman case–"A Catastrophic Success"–does not seem too strong to me for a hypothetical Japan that had gone full kana-ism. And here we get to the big difference between the Korean and the Japanese case. Professor Mair quoted a Japanese linguist above saying:

    I think the real issue is whether or not political leadership in China, Taiwan, Korea, and Japan will acknowledge and encourage the natural trend toward digraphia in their respective countries in a constructive way, or resist it in a reactionary way…

    Putting aside the mind-boggling comfort with cultural teleologies here ("natural", "reactionary"), I have trouble understanding what he/she could be referring to in the Japanese case (surely not that romaji will "naturally" encroach on kana?). Japanese literacy has been digraphic for at least over a 1000 years. If in error I beg your correction, but my understanding is that the Korean form of digraphy you observed is quite a bit younger in origin. Because hanja were mostly used to actually write in Chinese, Korean literacy in hanja for most of the country's history seems to me as (very roughly) akin to say, British literacy in Latin– centered around the ability to read and write a language which was foreign, yet nonetheless through massive borrowings on a continuum with one's own. I think the almost perfect similarity of the modern function of the Chinese element in Korean and Japanese tends to obscure the fact that the two cultures have very different histories of character usage. Literate Japanese have almost always experienced kanji as a way to write Japanese.

    Thus, while of course it would be possible to write entirely in kana–even entirely in hiragana–this would be a move towards an absolutely unprecedented monography. Certainly there have been many individuals throughout history who read (and wrote?) mostly only kana, but a flexible digraphy has nonetheless comfortably remained the standard throughout numerous dramatic expansions in literacy up to the present. Indeed, if functional digraphy is desired, Japan has long been proof-of-concept. Like Korea, with "certain cultural adjustments," Japan too could make the phonetic switch, but I submit that, when some form of the current system under question has been in continuous use to write the same language for a millennium, it is not surprising that the shift does not recommend itself.

RSS feed for comments on this post