Disappearing readings of Sinoglyphs: focus on Bo (–> Bai) Juyi / Haku Rakuten

« previous post | next post »

When I learned Mandarin half a century ago, it was a matter of faith, rectitude, and integrity that one should pronounce 說服 ("persuade") as shuìfú, not shuōfú, because when 說 is used with the meaning "convince; persuade", its pronunciation should be shuì, not shuō, which means "say; speak; explain", the more usual reading.  Now, however, in the PRC, according to my students from there, the pronunciation shuì basically no longer exists, not even when the character 說 is intended to mean "convince; persuade", and not even in many dictionaries.

說 can also be pronounced yuè, in which case it means "happy; delighted", and is the equivalent of 悦 (and compare my remarks on the equivalent meaning / reading of 樂 below).

In addition, 說 can also be pronounced tuō and means the same thing as 脱 ("to free; relieve").

My impression is that this change — and many others like it — was done by government fiat, and not through natural phonological evolution.  For example, 星期 ("week"), which I learned as xīngqí and is still pronounced that way in Taiwan, but which on the mainland is pronounced xīngqī.  There are many characters whose readings got shifted from other tones (especially second tone) to first tone.  When I first recognized this phenomenon, I referred to it as "language engineering".

The same thing has happened to hundreds of other characters.  Of course, however, many such sound changes have occurred naturally or through misreading.  A particularly interesting instance is 垃圾 ("garbage; rubbish; trash; waste; refuse; junk"), which in Taiwan is pronounced lèsè, but on the mainland is pronounced lājī.

From earlier (“to jumble; jumble; litter”). Reduplication derived from a root (*lap or *sap) meaning "untidy, unclean".

Second syllable of the Mainland Mandarin reading lājī is irregular, resulting from the incorrect association of the rare character in its obsolete form () with the usual pronunciation of its phonetic component ( ()).

Cognate with 邋遢 (lāta, “unclean, untidy, unkempt”) (> 邋邋遢遢, 邋裡邋遢邋里邋遢).

(source, which seems to draw from Victor H. Mair, "On 'Transformationists' (bianjia) and 'Jumbled Transformations' (laza bian): Two New Sources for the Study of 'Transformation Texts' (bianwen): With an Appendix on the Phonotactics of the Sinographic Script and the Reconstruction of Old Sinitic." In Alfredo Cadonna, ed., India, Tibet, China: Genesis and Aspects of Traditional Narrative. Orientalia Venetiana, VII. Firenze: Leo S. Olschki, 1999. Pp. 3-70."  pdf available here)

A final example of a polyphonic glyph that is losing one or more of its readings is 樂.  Here are the possibilities:

  1. yuè — music; tune; melody; song
  2. lè — joyful; happy; glad
  3. yào — to be fond of; to enjoy; to appreciate
  4. lào — used in place names
  5. luò — used in compounds

Here's part of what I wrote about the phonological vagaries of 樂 more than eight years ago:

The multiple pronunciations and meanings of 樂 lead to much mischief in the reading of Chinese texts, especially Literary Sinitic texts where there is a far higher percentage of monosyllabic words than in modern Sinitic languages.

In fact, one pronunciation of 樂, namely yào ("like; take delight in") seems to have dropped out of circulation on the Mainland altogether (I polled about 40 of my Mainland graduate students, and none of them knew it), though I learned it in Taiwan during the early 70s, and I think it is still alive there. And there's yet another pronunciation for 樂, viz., lào, which is used in place names — e.g., 樂亭 in Hebei and 樂陵 in Shandong — though I have no idea how locals and non-locals in China (and outside of China) actually pronounce those names nowadays.

Such are the delights / pleasures / joys (lèqù 樂趣 ["happy inclinations"]) of reading Chinese.


The loss of the yào reading is particularly troublesome for traditional texts.  A well-known passage from the Confucian Analects is 「知者樂水,仁者樂山。」《論語·雍也》– "The wise enjoy water, the humane enjoy mountains".  According to the traditional reading, this should be "zhìzhě yào shuǐ, rénzhě yào shān", which — except for some small differences in parsing — Google Translate (rather amazingly, because it's not designed for Literary Sintitic) gets right.  Here 樂 is read as yào because it's a transitive verb.  However, nearly all native speakers of Modern Standard Mandarin (MSM) from the PRC read this as "zhìzhě lè shuǐ, rénzhě lè shān", where the lè pronunciation, normally understood as an adjective or, more properly, a stative verb (which would make it intransitive), is forced to function on an ad hoc basis (for this literary / classical quotation) as a transitive verb.  Indeed, for all intents and purposes, the yào reading of 樂 has ceased to exist.

Now comes a scholarly matter that has bothered me for decades….

We know that the great Tang poet Bó Jūyì 白居易 (772–846) was influential in the development of Japanese literature, where he is better known by the on'yomi (Sinitic style) reading of his courtesy name as Haku Rakuten 白樂天.  That reading of 樂天 as Rakuten is tricky because, according to traditional literary / classical readings in Sinitic, it should be yàotiān 樂天, where yào indicates a transitive verb ("like; enjoy; take pleasure in"), whereas the lè of lètiān 樂天, which corresponds to Japanese raku of rakuten, is an adjectival / stative — i.e., intransitive — verb ("joyful; glad; happy").  This leads me to believe that the yào reading of 樂 had already dropped out of Japanese (or rather, perhaps, had never entered) Japanese by Bó Jūyì's 白居易 time (772–846), so they had to press "raku" into service in its stead.

After pondering this problem for decades, I've finally decided to make an effort to determine just what the Japanese readings for 樂 were in 白居易 / 白樂天's age.


for the yuè ("music") reading:

(BaxterSagart): /*[ŋ]ˤrawk/ (Zhengzhang): /*ŋraːwɢ/

for the lè ("joyful; happy; glad") reading:

(BaxterSagart): /*[r]ˤawk/ (Zhengzhang): /*raːwɢ/

for the yào (to be fond of; to enjoy; to appreciate") reading:

(BaxterSagart): /*[ŋ]ˤrawk-s/ (Zhengzhang): /*ŋraːwɢs/

My question, then, is why the Japanese chose "raku" [Middle Sinitic /lɑk̚/] (corresponding to MSM lè) for the on'yomi reading of the 樂 in 樂天 when, according to Classical / Literary Sinitic grammar and phonology, it should have been the Japanese equivalent of Middle Sinitic /ŋˠauH/ (corresponding to MSM yào) — if the latter reading were available in the poet's time?

Reply from Jim Unger:

There are only two on’yomi for the character:  raku and gaku.  My guess is that, very early on, gaku became so tightly associated with ‘music’ that raku was used for the ‘pleasure’ sense regardless of the grammatical status of the Chinese morpheme.  If that is correct, then this would be just a particular instance of the gradual divergence of Japanese kanbun usage from Chinese wenyan usage. 

From John Bentley:

Interestingly, the common readings for 楽 in Japanese are either gaku or raku. The reading based on yao is gyou  or gou, which appears to be rarely attested. For fun, I went through Morohashi, and looked at the 12 pages of attestations he has for compounds with 楽 (356 examples). There were a number of examples where the native Japanese reading is listed, but the vast majority of the readings were either gaku or raku. There were a few strange readings, like 楽正求 pronounced as gaiseikyu. Morohashi has this note under 楽天: le4 t'ien1 雲雀の別名. I wonder if the Japanese rendition of the poet's name was based on poetic considerations. At any rate, I didn't find any compounds where 楽 was read gyou or gou. The simplest answer may be that that reading, while theoretically available, wasn't in the Japanese lexicon, so the tradition defaulted to raku.

From John Whitman:

Japanese does have all three readings, and the most Sinographically sophisticated scholarship does recognize the relation between the reading corresponding to ŋˠauH > yào and the transitive interpretation ‘enjoy, take pleasure in’. But I don’t think that level of scholarship played a part in the establishment of Haku Rakuten as the reading of Haku Rakuten for 白樂天, as 白居易’s name in Japan. As you note, Bai Juyi was super-popular in Heian Japan, and the popular reading of the courtesy name simply took the most common reading of 樂, without thinking about the meaning of 樂天. (By the way, was it really supposed to mean ‘enjoy heaven’?).

It is a little like the vagaries of the various (mostly Latin, sometime Greek) names that were so popular in Upstate NY after the American Revolution. There is both a Cicero NY (founded 1790) and a Tully NY (1803). Surely very few know that they refer to the same person, as philosophers remind us. Tully was probably the better-established name among anglophones who had actually been taught some Latin, at the time. It is also more famous now, due to the ubiquitous Tullyburger.

You can find the three Japanese readings for 樂 on Wiktionary. The Wiktionary entry does correctly relate the readings derived from MC (呉音 ギョウ < ケウ, 漢音 ゴウ < ガウ) to transitive konomu ‘likes’, which has the possible kun spelling 楽む (although this probably won’t come up in your application for writing Japanese). But you’ll note that the much more common transitive kun reading tanosimu 楽しむ ‘enjoys’ is associated with the onyomi raku, same as Bai Juyi’s name – reflecting the same spread of the most common reading. Japanese Sinologists were aware of the history of the readings of 樂, but high school kanbun teachers were probably not in most cases.

For evidence of this, you have to go to Morohashi. Page 506, vol. 6 of the 大漢和辞典 gives all three readings, as well as a fourth reading レウ which is based on the 集韻 fanqie 力照切. I’m not sure but I think this probably reflects the use of 樂 as a variant form of 療, also mentioned in Wiktionary.

The readings ギョウ < ケウ andゴウ < ガウ are rare, and scholarly. You can find one on p. 517 of Morohashi vol, 6, where the quotation 樂此不疲 from the Hou Han Shu gets the onyomi reading カウシフヒ (I think there is no dakuten on the カウ but I’d need a magnifying glass to make sure; if so, this shows how Morohashi really cites his examples from actual texts: the kan-on should be ガウ, but the カウ spelling probably reflects some (Edo?) scholar assimilating this to the absence of dakuon in kan-on oral velars.

If you read through the examples on this page, you’ll find that the Lunyu citations are all given 読み下し, that is kun readings, with no on-yomi. For example 樂而不淫 is given the reading タノシミテインセズ. This is because these are total warhorse passages that everyone learns to read as a yomikudashi in kun. I would bet my bottom dollar that if students in middle school or high school are made to give an on-yomi for this and similar passages, they read the first character as raku. But I’ll check with people who actually did kanbun in high school to make sure and get back to you if any of them respond.

Did this thread originate with a discussion of the source of the Japanese company name Rakuten?* If so, the best discussion of that that I have been able to find is here. Many sites on the web, including this one, claim that the source, or one source, was the name given by the pre-Edo warlord Oda Nobunaga (also of video game fame) to his economic policy, 楽市楽座. But this doesn’t make sense because there is no 天 in it. The page cited above suggests that the company founder, Hiroshi Mikitani, wanted to fuse the Nobunaga reference with the term 楽天, as in 楽天家, 楽天主義, etc. That makes sense. I always thought that the 楽天 forms were pretty recent Japanese coinages, but Baidu gives a bunch of very old Chinese citations in the basic sense ‘happy go lucky’.**

[*VHM:  No, I dreamed this inquiry up myself because of being puzzled by the current Chinese reading of Letian instead of the traditional one, Yaotian.]

**That "happy go lucky" interpretation for 楽天 would go well with the poet's original name, 居易 ("dwell at ease"), which also has deep roots:

Lǐjì 礼記 (Record of Rites)  Zhōngyōng 中庸 (Doctrine of the Mean)

Jūnzǐ jū yì yǐ qí mìng, xiǎorén xíng xiǎn ér jiǎoxìng 君子居易以俟命、小人行険而僥倖 ("The gentleman dwells at ease and awaits fortune; the petty person courts danger and risks his luck.")

Zihan Guo reminded me of another relevant quotation from the Analects:

"Zǐ yuē:`Zhīzhī zhě bùrú hào zhī zhě, hào zhī zhě bùrú yào zhī zhě.' 子曰:「知之者不如好之者,好之者不如樂之者。」" Lúnyǔ · Yōngyě 論語·雍也

"The Master said, 'They who know the truth are not equal to those who love it, and they who love it are not equal to those who delight in it.'" (James Legge, tr.)

Also from the slightly later Mencius (372-289 BC):

"Yǐ dà shì xiǎo zhě, yào tiān zhě yě; yǐ xiǎo shì dà zhě, wèi tiān zhě yě.  Yào tiān zhě bǎo tiānxià, wèi tiān zhě bǎo qí guó.  以大事小者,樂天者也;以小事大者,畏天者也。樂天者保天下,畏天者保其國。Mèngzǐ · Liáng huì wáng, xià 孟子·梁惠王下

"He who with a great State serves a small one delights in Heaven.  He who with a small State serves a large one stands in awe of Heaven.  He who delights in Heaven will affect with his love and protection the whole kingdom.  He who stands in awe of Heaven will affect with his love and protection his own kingdom." (James Legge, tr.)

In MSM, the expression 樂天 ("enjoys / likes / is fond of / takes pleasure in / delights in heaven") describes people with a positive mentality, i.e., optimists.  樂天 is a transitive verb-object construction, so the traditional reading would be yàotiān, but virtually everybody I know pronounces it lètiān.

From Bob Ramsey: 

The poet is just as loved in Korea, where he is also called by that same name (today!) read Baeg Nagcheon 백낙천.  (Before recent times, the name was always written in Chinese characters, of course.) And, you know, I'm pretty sure the Koreans did not get that reading from Japan….

Dwell at ease, my friends, and take pleasure in heaven.  Even if we lose a few readings, it's not the end of the world.


Selected readings

[Thanks to Bryan Lowe, Timothy Billings, Bruce Batten, and Cynthea Bogel]


  1. Chris Button said,

    September 18, 2022 @ 9:52 pm

    I’ve always found the onset ŋr- in Old Chinese somewhat problematic (a little like the rather rare -ŋʔ coda—it’s there, but it’s awkward). Interestingly, Old Burmese has kr-, pr-, mr- but no ŋr- (despite some misleading orthography occasionally suggesting otherwise).

  2. Chris Button said,

    September 18, 2022 @ 10:25 pm

    … as well as a fourth reading レウ which is based on the 集韻 fanqie 力照切. I’m not sure but I think this probably reflects the use of 樂 as a variant form of 療, also mentioned in Wiktionary

    That would make good sense

  3. Chris Button said,

    September 19, 2022 @ 6:22 am

    It occurs to me that the meaning of 療 “cure” could be etymologically derived from 樂 “joy” via the -s coda: “make joyful” meaning “cure”

  4. Victor Mair said,

    September 19, 2022 @ 4:03 pm

    From Haun Saussy:

    As far as I know, Pound never said anything about Bo Juyi. But Fenollosa’s notes on Chinese poetry from Mori’s lectures, unpublished so far, give this (the transcription from the Yale Collection of Fenollosa notebooks is by Timothy Billings):

    About same time as R. lived Hakurakutei [Bai Juyi]. He too / wanted to base his style on Ri To. But what point to seize? / R. had monopolized boldness of thought. So he went to opposite / extreme — made it his special art to express things in the most / ordinary way, and yet make it poetical. All can understand it. / All the poems have this character. It is said that R. used to / read to his grandmother, and ask if she understood. If not, he / corrected. These are the great To men.

    Hakurakutei is of course 白樂天. Ri To = 李、杜. To = 唐. I think in the story about the grandmother “R.” is a mistake for “H” (Hakurakutei). Fenollosa was taking notes as Mori spoke and Ariga gave simultaneous translation, so mistakes are inevitable.

  5. Victor Mair said,

    September 19, 2022 @ 4:06 pm

    From David Lurie:

    Your quoted correspondents know more about the history of Japanese "Sinoxenic readings" than I do, but John Whitman's points about the narrow scholarly quality of the more precise on'yomi and the broad applicability of the more familiar ones strike me as particularly valuable. For 樂, the Kanchi'in version 観智院本 of the Ruijū myōgishō 類聚名義抄 dictionary, which dates to the late Heian or early Kamakura period, lists raku as a "Japanese on'yomi" (和音), which essentially means a familiar and widely used on'yomi unlinked to the scholarly pedigree implied, for example, by citation of fanqie spellings. I'm sure John is correct that the default quality of this reading is what leads to the Haku Rakuten pronunciation of the poet's name.

  6. B.Ma said,

    September 20, 2022 @ 1:11 am

    I'm pretty sure 說 meaning convince is still pronounced as seoi3 in Hong Kong; I heard it on HK TV recently and I think one of my friends said it too.

    I wonder if Cantonese speakers on the mainland have changed their pronunciation to syut3 due to influence from the "wrong" Mandarin pronunciation?

    If the Mandarin pronunciation of shuì has died out on the mainland, is knowing it now regarded as "trivia" that one can use to demonstrate one's smartness? Similar to people who argue the plural of octopus in English should be octopodes.

  7. Victor Mair said,

    September 20, 2022 @ 8:23 pm

    Michelle Frey, "What is the Plural of Octopus? Octopi vs octopuses vs octopodes", Ocean Conservancy (2/1/22):


    “Octopodes” stems from the belief that because octopus is originally Greek, it should have a Greek ending. This term might be technically correct, but it is the least-used incorrect form of the word for more than one octopus. Using “octopodes” might cause more confusion than it’s worth.


    That's quite different from the situation regarding 說服 ("persuade") where, until quite recently, everybody speaking Modern Standard Mandarin read it as shuìfú, not shuōfú. Indeed, it was my students from the PRC who called this replacement of shuìfú by shuōfú to my attention.

  8. Chris Button said,

    September 21, 2022 @ 1:10 pm

    The shuì reading is nice too from a historical perspective. It contains both the “transitivizing/causativizing” s- prefix and -s suffix.

  9. DMT said,

    September 21, 2022 @ 7:52 pm

    >"until quite recently, everybody speaking Modern Standard Mandarin read it as shuìfú, not shuōfú"

    I discussed this word with friends in Mainland China about 20 years ago; they were aware that shuìfú was "correct," but they told me many people at the time already pronounced it as shuōfú.

  10. Vampyricon said,

    September 22, 2022 @ 4:45 pm

    Can confirm that 說 in the sense of "persuade, convince" is still pronounced seoi3 in HK Cantonese. Sometimes it's even used monosyllabically.

  11. Terpomo said,

    September 22, 2022 @ 5:54 pm

    This leads me to wonder- are MSM speakers taught about distinct readings of the same character that have become homophones in MSM by normal sound change (like 近, 重, 後, 易) but which may be retained in some other lects? Even if they don't say them any differently, being aware of the distinction might help them interpret traditional commentaries that talk about characters' readings in terms of fanqie or the four tones of Middle Chinese.

RSS feed for comments on this post